SCA2016-004 1/12

RECOMMENDED PROCEDURE FOR DETERMINATION

OF RELATIVE PERMEABILITIES
Roland Lenormand and Guillaume Lenormand, Cydarex, France

This paper was prepared for presentation at the International Symposium of the Society of Core
Analysts held in Snowmass, Colorado, USA, 21-26 August 2016

ABSTRACT
This paper discusses the basic mechanisms involved during determination of oil/water
imbibition relative permeabilities (Kr).

One-step experiments (only one injection pressure or flow-rate) generally referred to as

JBN experiments must be discarded even if experiments are interpreted with Pc curves.
Their interpretation is based on transient flow dominated by viscous fingering and/or
channeling, which does not represent "true" pore-scale relative permeability.

A more accurate UnSteady-State (USS) experiment requires 7 to 10 steps. After the first
step, saturations become more uniform and flows are then controlled by local Kr. The
range of saturation is controlled by the balance between viscous and capillary forces and
works only for the negative part of the Pc curve. The experiment is difficult to design if
the Pc curve is not well-known.

Simultaneous injection of oil and water (steady-state, SS) allows a more precise control of
saturations. When two fluids are injected at high flow rates, saturation is close to being
uniform on a large part of the plug, and its value is controlled by the ratio of viscous
forces in the two fluids. This method can be used for any type of wettability.

Based on these observations, the recommended procedure is a compromise between SS

and USS methods. We also show that laboratory fluid velocities are much higher than
reservoir velocities and discuss how to deal with this issue.

INTRODUCTION
The determination of relative permeabilities by displacements is still a subject of much
discussion in oil companies. There is always a debate between Steady-State (SS) and
UnSteady-State (USS) methods. The purpose of this paper is to explain the physical
mechanisms that govern these displacements, and how they affect the determination of
relative permeabilities.

For oil and water displacements, the following definitions are used, independently of
wettability: drainage is a displacement leading to an increase of oil saturation, and
imbibition to an increase of water saturation. We will use oil/water imbibition, the most
 SCA2016-004 2/12

common case for relative permeabilities. Results can easily be extended to other types of
displacements, such as gas drainage.

Kr curves are defined and calculated at local scale, the scale of a Representative
Elementary Volume where the two-phase Darcy's laws are written:
Qw K Krw ∂Pw Qo K Kro ∂Po
=− ; =− (1)
A µw ∂x A µo ∂x

where, Q is flow rate, K absolute permeability, Kr relative permeability, µ viscosity, P

pressure and x the distance along the sample from the injection face. Indices are "w" for
water and "o" for oil. Capillary pressure is defined by Pc = Po − Pw .

A laboratory experiment is not a "small scale" representation of the reservoir. Especially,

the residual oil saturation (Sor) obtained in laboratory depends on laboratory conditions,
sample length, flow rates, etc. and differs from the recovery at reservoir scale, controlled
by other parameters (geological structure, well implementation, etc.). The purpose of a
laboratory experiment is to determine a full curve of relative permeabilities (Kr) as
functions of water saturation (Sw) in a very large range of saturation that covers or even
exceeds the saturations encountered in the reservoir.

For a given type of displacement (drainage or imbibition), the Kr curves are assumed to
be independent of fluid velocities. This means that the viscous coupling between the
fluids is negligible, as if the two fluids were flowing in separate channels (see discussion
by Ayub and Bentsen [1]). With this assumption, the Kr curves are the same for SS, USS
and other types of displacements such as centrifuge experiments. This assumption is in
contradiction with the discontinuous flows observed at high flow rates both in
micromodel experiments and Digital Rock Physics simulations. It is also often observed
that Pc curves determined by history matching differ from Pc curves measured by porous
plate or centrifuge methods. Lackner et al. [2] attribute this difference to "dynamic
capillary pressures" and recommend local pressure measurements using semi-permeable
membranes. We will not discuss this point here and assume that both Kr and Pc curves do
not depend on fluid velocity.

We will start by discarding the JBN method, in which the information obtained is limited
and not accurate, especially in terms of Sor. Then we will analyze the USS and SS
methods in terms of balance between viscous forces and capillary forces. Finally we will
compare laboratory experiments to numerical simulations at reservoir scale to discuss the
problem of fluid velocities and the important notion of Sor.

ONE FLUID INJECTED –ONE STEP (JBN)

This simple experiment of water displacing oil in a sample at irreducible saturation Swi, is
usually referred to as JBN, from the name of the authors of the paper describing the Kr
 SCA2016-004 3/12

analytical calculation [3]. This method is still often used, because it is quick and cheap,
but it presents several drawbacks.
1

Local Krw
0.8 Local Kro

relative permeability
Krw JBN
0.6 Kro JBN

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
water saturation

Figure 1 - (from Kokkedee et al) Figure 2 - (from Kokkedee et al). Figure 3 - Kr determination
Saturation profile along the Analytical and numerical Kro. assumed on a heterogeneous
plug. sample (from Fenwick et al.)
More than 20 years ago, Kokkedee et al. [4] showed that there was always an important
error on Sor when the JBN method was used. Whatever the reservoir type, Sor obtained
by numerical history matching with capillary pressures is always smaller than the JBN
Sor by 15% to 30%. The explanation is illustrated on Figure 1. The JBN calculation
assumes a uniform saturation over the sample; however the profile is not uniform and
drops near the outlet due to capillary effects. Although the average, derived from the
effluent balance, is the same, the "local" saturation near the inlet face is larger than the
average saturation. Figure 2 shows the effect on the oil Kr curve.

Another drawback is channeling. In 1988, Mohanty and Miller [5] wrote: "Results
illustrate that the early part of the JBN method relative permeability is dominated by
fingering and heterogeneity effects. But the later part (> 1PV) represents the relative
permeability of the end-face saturation". This result has been confirmed and illustrated by
numerical simulations (Fenwick et al. [6]). In their study, the sample is represented as a 2-
Dimensional heterogeneous permeability field with uniform Corey-shapes Kr curves for
all the grids (dashed lines in Figure 3). The displacement is computed using a streamlines
method and the Kr curves are determined from the production and pressure drop, as in a
real experiment (solid lines). At low saturations, Krw is larger than the local value
because water is flowing in channels more permeable than the average. At the end of the
experiment, Kr are close to the local values. The beginning of displacement is also prone
to viscous fingering with an adverse mobility ratio (see for instance Sarma et al. [7]). Due
to all these problems the one-step injection of water (JBN type experiment) is no longer
recommended, even if numerical simulations are used with Pc curves. Multi-step
experiments, either SS or USS, provide more accurate results.

MULTI-STEP EXPERIMENTS
For both USS and SS multi-step experiments, the relative permeabilities are determined
by history matching of the transient and stabilized parts of differential pressure and oil
production. The numerical simulation takes into account a Pc curve, either measured or
adjusted from the experimental results together with Kr. As for JBN, the heterogeneity
plays a role at the beginning of the first displacement. At the end of the first step and all
 SCA2016-004 4/12

the further steps the sample is assumed uniformly saturated with the two fluids and that
the Kr are not affected by channeling and fingering.

Using numerical optimization does not provide "points" Kr values as for analytical
calculations, but an analytical curve (Corey, LET, etc.) determined by minimizing the
mean square root difference between experimental and calculated parameters during both
transient and stabilized regimes. The most accurate determination of the Kr curves is
when average saturations are numerous and regularly spaced.
Physical mechanisms during USS multi-step displacements
In imbibition, only water is injected, starting at a low flow rate and allowing for
stabilization of oil production and differential pressure. Then the water rate is increased,
until stability is reached and the process is repeated for 5 to 10 steps. At each step the
average saturation increases. We will show that in a USS experiment, average saturation
at each step is controlled by the balance between viscous forces and capillary forces.
Another important result is that the differential pressure is equal to capillary pressure at
entrance when stabilization is reached.

For illustration on a general case, the Pc curve is assumed to have a positive and negative
part (mixed wettability). For any step, when stabilization is reached, the water and oil
pressures along the samples are schematically displayed in Figure 4a:
• At outlet, the capillary pressure is assumed to be zero; consequently both oil and water
are equal to the outlet imposed pressure, taken as reference (P = 0).
• The pressure in oil is uniform since oil is no longer flowing when equilibrium is
reached (no gravity).

a) b) c)

Figure 4 - USS displacement: a) pressures profiles at equilibrium (water blue, oil red), arrows represent
the capillary pressure at several locations along the sample, b) determination of corresponding saturations
from the Pc curve, c) saturation profile.

• Water is flowing through the sample and since the outlet pressure is 0, the pressure in
water is positive everywhere along the sample (a fluid flows from higher to lower
pressure). The pressure measured in the inlet end-piece is the highest pressure
between the water and oil pressures inside the sample at the entrance, equal to the
pressure in water (this point is discussed in [8]) Since Pc = Poil - Pwater, Pc is negative
at any location along the sample (arrows in Figure 4a). Consequently, the difference
 SCA2016-004 5/12

of pressures measured between the inlet and outlet end pieces is equal to the opposite
of the capillary pressure at inlet face (Figure 4a). This property is the principle of
determination of the Pc curve by the semi-dynamic method [8].
• From the value of Pc at any location (B, C, D), the corresponding saturation is
imposed by the capillary pressure curve (Figure 4b), leading to the saturation profile
(Figure 4c).
• At outlet Pc=0 and the corresponding saturation S(D) is equal to the saturation at Pc=0
(where the curve cuts the saturation axis). For an "oil-wet" sample, with totally
negative Pc curve, the saturation at outlet is equal to Swi.

Figure 5 shows the flow rate increased by a factor 2. Most of pressures are now in the
near vertical part of the Pc curve (Figure 5a and b). Saturation is more uniform on a large
part of the sample with a smaller capillary end effect. Average saturation of water has
increased and during the experiment a production of oil has been observed.

a) b) c)

Figure 5 - USS displacement: similar to Figure 4, but with a higher flow rate a) pressure profiles in solid
line, previous profiles in dashed lines, b) higher local Pc leads to higher local water saturations, c)
saturation profile with the higher flow rate (solid line) showing a decrease of the extend of the capillary end
effect.

a) b) c)

Figure 6 - Pressure and saturation profiles for a water-wet sample. Since Pc is negative, the corresponding
saturation is always equal to 1-Sor.
For a water-wet sample (Figure 6), the water pressure profile may defer from the previous
case since Kr are different, but the overall shape is similar with positive pressures
everywhere in water that is flowing. Capillary pressure along the sample is always
negative and for all locations, the local saturation is equal to the maximum value (1-Sor).
 SCA2016-004 6/12

When flow rate is increased, saturations do not change since they are at their maximum
values and there is no production after the first step.

The main conclusion is that capillary pressure is always negative along the sample.
Consequently, local saturations are determined from the negative part of the Pc curve. In
drainage it is the opposite.
Numerical example USS
We now illustrate the previous explanation using a numerical simulation with the
commercial software CYDAR (http://www.cydarex.fr). Sample, fluid properties and Kr
are given in Table 1 and flow rates in Table 2.
Table 1 - Sample and fluid properties for 1
numerical simulations
0.8

average water saturation

type of displacement imbibition
0.6
disposition horizontal
length cm 8 0.4
Diameter cm 4
0.2
Base permeability mD 100
porosity frac 0.3 0
water viscosity cP 1 0 5 10 15 20
Time (hour)
25 30 35

water density g/cm3 1

oil viscosity cP 5
Figure 7 - USS experiment: Water average saturation.
oil density g/cm3 1
initial Sw frac 0.2
final Sw frac 0.8
Krw_max frac 0.5
Kro_max frac 0.5
Corey exponent water 3 Qw
Corey exponent oil 3

Table 2 - Flow rates during the USS

experiment
duration 5 5 5 5 5 5 5
(hour) Figure 8 - USS experiment: saturation profiles
rate cc/h 2 20 100 250 450 500 1000

capillary pressure
Figure 7 shows the average water saturation as
2 <Sw>
Sw face
function of time. Figure 8 shows saturation
1 profiles at end of each step. For each step there
is an increase of average water saturation and
pressure (bar)

consequently additional production of oil.

-1

-2
Figure 9 shows capillary pressure at the inlet
-3
0 0.2 0.4
face at each stabilized step as function of the
0.6 0.8 1

average saturation and local saturation. As

Water saturation

Figure 9 - USS simulation: inlet face capillary explained previously, only saturations
pressure versus average water saturations corresponding to Pc<0 are obtained.
and inlet face saturations;
 SCA2016-004 7/12

Physical mechanisms during SS multi-step displacements

Oil and water are injected together starting with 100 % of oil at Swi. Then several flow
rates (5 to 10) are used by increasing the fraction of water until 100% of water is injected.
The displacement mechanisms are more complicated than for the USS case since there are
now three types of forces: viscous forces in each fluid and capillary forces.

At the beginning of the experiment, the flow rate is much higher in oil than in water (low
water fractional flow) and Pc>0 since pressure in oil is higher than pressure in water
(Figure 10a). The local saturations correspond to the positive part of the Pc curve (Figure
10b and 10c). For higher water fractional flow, Pc is negative and local saturations
correspond to the negative part of the Pc curve (Figure 11).

a) b) c)

Figure 10 - SS displacement at low water fractional flow: pressure is higher in oil and capillary pressure is
positive

a) b) c)

Figure 11 - SS displacement at high water fractional flow: pressure is higher in water and capillary
pressure is negative
Numerical example SS
Experiments are simulated using the same fluids and sample properties as for the USS
case (Table 1) with 3 total flow rates Qt = 2, 20, and 1000 cc/h. Saturation profiles at end
of experiments are shown in Figure 12.

Figure 13 and Figure 14 display the fractional flow as function of average saturation. The
numerical fractional flow Qw/(Qw+Qo) is obtained from the simulations when
stabilization is reached and the analytical values are derived from Darcy's law with the
assumption of Pc=0:
 SCA2016-004 8/12

Krw ⎛ Krw Kro ⎞

fw = ⎜ + ⎟ (2)
µw ⎝ µw µo ⎠
At high total flow rate, Qt = 1000 cc/h
Qt = 1000 cc/h
(Figure 13), analytical and numerical
Qt = 20 cc/h fractional flows are very close and well-
Qt = 2 cc/h spaced saturations can be determined over
all the range of saturations by imposing the
fractional flow, independently of the
capillary pressure (negligible forces). For
lower total flow rate, capillary forces are
acting and the numerical fractional flow is
Figure 12 - saturations profiles at end of experiment no longer close to the analytical one
for 3 total flow rates. (Figure 14).
In addition, for the lowest total flow rate case (2 cc/h), the range of saturation decreases
and the main value is centered around 0.5 where Pc =0. For Qt = 20 cc/h, the average
saturation is only in the range 0.3 – 0.6. In the limit of very low flow rate (dashed line in
Figure 14), viscous forces are negligible and the final saturation is uniform and equal to
0.5 (spontaneous imbibition).

Figure 13 - High total flow rate: Analytical and Figure 14 - Analytical and numerical fractional
numerical fractional flows are very close and flows as function of average saturation for several
saturations can be determined by imposing the total flow rates.
fractional flow.

RESERVOIR SCALE DISPLACEMENT MECHANISMS

We will present results of numerical simulation at reservoir scale to illustrate the
applications and limitations of laboratory experiments. The reservoir is schematically
represented by a 200 meters 1-D medium with same permeability and porosity than the
laboratory sample. For Kr properties, we have used 2 cases:
• Low mobility, same as the laboratory simulations with Corey exponents α = 3;
• High mobility, same other properties but with Corey exponents α = 2.
 SCA2016-004 9/12

For field velocity, we have used the rule of thumb of 1 foot/day and chosen 0.7 foot/day,
corresponding to the laboratory injection at 2 cc/h (Table 1). Oil production during 30
years is shown in Figure 15. As expected, production is faster for high mobility. An
important result is that after 30 years, even if oil rate is very low, there is still an important
pressure gradient in oil along the reservoir, close to the gradient in water and the capillary
pressure is higher than -1.2 bar everywhere, even for the high mobility case (Figure 16).
From Pc profiles, we have calculated saturation and corresponding Kroil profiles (Figure
17). Another important point is that Kroil is always larger than 10-4 (dashed red line)
except very close to the injection well.

Table 3 - Equivalence between laboratory total flow rate

and front velocity
cc/h 2 20 100 500 1000
foot/day 0.7 6.5 65 162 323

0.8

oil production
0.7

0.6

0.5
oil production (OOIP)

high mobility
0.4 low mobility

0.3

0.2

0.1

0
Figure 16 - Numerical simulation for a 1-D
0 5 10 15
time (year)
20 25 30
reservoir after 30 years of production: water
and oil pressure profiles and capillary pressure
Figure 15 - Oil production during 30 years. for 2 values of oil mobility.

Figure 17 - Numerical simulation for a 1-D reservoir

after 30 years of production: water saturation and Figure 18 - Range of saturations obtained during
Kro profiles for 2 values of oil mobility. the USS and SS displacements performed at
various flow rates.
 SCA2016-004 10/12

DISCUSSION
We will first present the advantages and drawbacks of SS and USS methods without
focusing on the flow rate. We will then discuss how to deal with the fact that reservoir
velocities are much lower than designed front velocity. We will finally discuss the notion
of residual oil saturation, laboratory values compared to reservoir final recovery.
One fluid injected (USS)
The main drawback is that there is no possibility to have saturations in the positive part of
the Pc curve and this method cannot be used for water-wet cases. In addition, when the
negative Pc presents a flat plateau, it is difficult to adjust the flow rates to have regularly-
distributed saturations. This point is the main limitation of the USS multi-rate method.
Two fluids injected (SS)
Simultaneous injection of oil and water (SS) allows a precise control of saturations. When
two fluids are injected at high flow rate, saturation is close to be uniform on a large part of
the plug, and its value is mainly controlled by the ratio of viscous forces in the two fluids.
This method can be used for any type of wettability. At lower flow rates, the range of
saturation is limited (Figure 18).
How to avoid high flow rates during laboratory experiments?
The first answer is to increase the
length of the core. Horizontal cores do
not exceed 8-10 cm and the standard
solution was to use composite cores,
several cores put together in the same
coreholder. Now, in-situ saturation
monitoring shows that there is a huge
discontinuity at each contact (Figure
19), and composite cores should be
avoided. A better compromise is to take Figure 19 - In-situ saturation monitoring along a
vertical plugs with length around 20 composite core showing the discontinuity at each
cm. interface.

What is the difference with reservoir? If a 8 cm slice of the reservoir is considered as a

plug, the main difference with laboratory is that the condition at outlet is not Pc=0 (this
condition is only imposed at the producing well) and local capillary pressures are much
higher in the reservoir (see pressure profiles in Figure 16). Semi-permeable porous plates
can be used to increase the Pc inside the sample. However, due to the huge pressure drop
through the porous plates and their possible partial invasion, it is not possible to derive
directly the Kr from standard porous plate experiments (Lenormand et al. [9]). Brown in
1951 [10] used a modified Hassler core holder with ceramics plates to measure "dynamic
capillary pressures" and to compare them to "static" values, but there was no result for
relative permeability. Oil was injected and produced through the ceramic plates and gas
through grooves at the contact between ceramics and the sample. A similar equipment
was patented by Rose in 1985 [11] for Kr measurement. More recently, Egermann and
 SCA2016-004 11/12

Fleury [12] have described an apparatus to measure Kr behind the porous plates, but their
equipment does not allow an outlet boundary condition different from Pc = 0. Centrifuge
experiments could be a solution to allow high capillary pressure with low flow rate.
However, Kr measurements with centrifuge are challenging (Bauget et al. [13]).
Recommended procedure
We will give an example using our numerical simulations. The lower saturations (Pc>0)
can be reached only using the Steady-State method. The range of saturations as function
of total flow rate is displayed in Figure 18. The highest flow rate can cover all ranges of
saturation but a flow rate of 20 cc/h is preferred in order to approximate reservoir
velocities. However, this rate limits the final saturation to 0.6. Afterwards a one-fluid
injection (USS) is performed with several steps to estimate the negative part of the Pc
curve, giving information on the wettability.
Asymptotic values for Sor
Our last point of discussion concerns the residual oil saturation (Sor). When using an
analytical function for Kr representation during numerical simulation (Corey or LET), the
asymptotic Sor value ( Pc → − ∞ ) is often equal to zero. For instance, the result presented
by Kokkedee (Figure 2) shows a tendency to reach asymptotic zero Sor value. This result
is not in contradiction with physical mechanisms: when samples are not strongly water-
wet, oil can flow very slowly on the surface of the solid up to very low saturation when
pressure is increased. However, this asymptotic value is not the final value reached in
reservoir production because capillary pressures are always limited, due to the low
mobility of oil, as shown in Figure 16.

CONCLUSIONS
• One-step experiments, generally referred to as JBN experiments, must be avoided
even if they are interpreted with Pc curves. Their interpretation is based on transient
flow dominated by viscous fingering and/or channeling, which does not represent
"true" pore-scale relative permeability.
• A more accurate experiment requires 7 to 10 steps: saturations become more uniform
after the first step, and flows are then controlled by local Kr.
• Capillary end effect is not an artifact that should be removed but rather an advantage
that allows exploration of all the range of saturations. Conducting Kr (or EOR)
experiments at maximum speed as described in [14] is not a recommended solution.
However, with standard displacement experiments, either SS or USS, the balance
between capillary and viscous forces requires fluid velocities to be much higher than
in reservoirs. An ideal experiment should be able to provide high capillary pressure
with low flow rate. Solutions should be investigated using porous plate or centrifuge
methods.
• Kr determination during experiments with capillary effects is only possible with
numerical simulators. Numerical simulators are also useful to design the experiments.
• When a single fluid is injected (unsteady-state, USS), the range of saturation is
controlled by the balance between viscous and capillary forces. Consequently, the
experiment is difficult to design if the Pc curve is not well-known.
 SCA2016-004 12/12

• Steady-state method allows a precise control of saturations. When two fluids are
injected at high flow rates, saturation is close to be uniform on a large part of the plug,
and its value is controlled by the ratio of viscous forces in the two fluids. This method
can be used for any type of wettability. At low flow rate, the range of saturation is
limited to a small interval around the saturation where Pc=0.
• Based on these observations, the recommended procedure is a compromise between
SS and USS methods.
• We finally show how the Sor measured in laboratory by history matching may defer
from the true reservoir Sor.

ACKNOWLEDGEMENT
We want to thank Jos Maas and Karl Sigurd Årland for providing authorizations to
reproduce their figures and Fabrice Bauget for useful comments on the manuscript.

REFERENCES
1. Ayub, M. and R.G. Bentsen, "Interfacial viscous coupling: a myth or reality?", (1999), J. Petroleum
Science and Engineering, Vol 23.
2. Lackner, A.S., O. Torsaeter, "Phase pressure measurements: simultaneous and direct derivation of
relative permeability and dynamic capillary pressure", (2005), SCA2005-05.
3. Johnson, E.F., D.P. Bossler, and V.O. Naumann, “Calculation of Relative Permeability from
Displacement Experiments”, Transactions AIME, 216, 1959, 370-372.
4. Kokkedee, J.A, W.Boom, A.M. Frens and J.G. Maas, "Improved Special Core Analysis: Scope for a
reduced residual oil saturation", (1996), SCA9601.
5. Mohanty, K.K. and A.E. Miller, "Factors influencing unsteady relative permeability of a mixed-wet
reservoir", (1988), SPE18292.
6. Fenwick, D., N. Doerler, and R. Lenormand, "The effect of heterogeneity on unsteady-state
displacements", (2000), SCA2000-30.
7. Sarma, H.K., B.B. Maini and G. Allen, "Effect of viscous instability on unsteady-state relative
permeability", (1992), Revue Institut Français du Pétrole, Vol. 47, Nov-Dec.
8. Lombard J.-M., P. Egermann and R. Lenormand, "Measurement of Capillary Pressure Curves at
Reservoir Conditions", (2002), SCA 2002-09.
9. Lenormand R., P. Delaplace and P. Schmitz, "Can we really measure the relative permeabilities using
the micropore membrane method", (1996), SCA9637.
10. Brown H. W., "Capillary pressure investigations", Petroleum Transactions, AIME, vol. 192, (1951),
also SPE-951067-G.
11. Rose W. D., "Apparatus and procedure for relative permeability measurements", US patent number
4,506,542, 1985.
12. M. Fleury, P. Poulain and P. Egermann, "A new approach to derive relative permeability data while
measuring resistivity index, SCA2005-04.
13. F. Bauget, S. Gautier, R. Lenormand and A. Samouillet, "Gas-liquid relative permeabilities from one-
step and multi-step centrifuge experiments", (2012), SCA2012-13.
14. Masalmeh, S.K, "Determination of waterflooding residual oil saturation for mixed to oil-wet carbonate
reservoir and its impact on EOR", (2013), SPE-165981-MS.
 SCA2016-001 1/12

A NOVEL LABORATORY METHOD FOR DETERMINING

CAPILLARY PRESSURE AND WETTABILITY WHILE
MEASURING STEADY-STATE RELATIVE
PERMEABILITY
Robin Gupta, Daniel Maloney, David Laverick, Robert Longoria, Larry Poore and Jeff
Spitzenberger, ExxonMobil Upstream Research Company

This paper was prepared for presentation at the International Symposium of the Society of Core
Analysts held in Snowmass, Colorado, USA, 21-26 August 2016

ABSTRACT
A novel laboratory method was developed that obtains relative permeability (kr),
capillary pressure (Pc), and wettability characteristics from a single reservoir-condition
steady-state coreflood test. The method uses a modified inlet end-piece that isolates
injection phases so that the difference between injection phase entry pressures is
measured. Several inlet end-piece designs and related instrumentation are developed that
can isolate and precisely capture the pressure difference of injection phases at the inlet.
A method is developed to extract the capillary pressure from the difference in flowing
injection phase pressures, which has both capillary and viscous pressure contributions.
For this method, injected phase pressures are measured for several total flow rates for
each steady-state fractional flow. This process enables quantification of Pc and kr at each
steady-state fractional flow condition. Rock wettability can be estimated from such
measurements using either USBM or Amott-Harvey methods when the steady-state test
includes both primary imbibition and secondary drainage cycles. Measured steady-state
live-fluid coreflood data is used to illustrate this method of gaining both Pc and kr
functions from a single test. An excellent match was obtained between Pc quantified by
the centrifuge method and the new method.

INTRODUCTION
Capillary pressure (Pc), relative permeability (kr) and wettability are key special core
analysis (SCAL) characteristics used for reservoir performance predictions. Common
practice is to measure these characteristics independently via several tests. Ideally, one
would like to obtain all of these characteristics for a rock and fluid system from one test,
saving time and cost. By doing so, a number of technical issues are avoided, such as
mismatched data sets, anomalies from using different samples, test conditions, and fluids
in the various tests, and propagation of errors from combining results from different tests.
When a sample (core plug or composite of core plugs) is mounted for a steady-state
coreflood test, it is placed between inlet and outlet flow distribution end-pieces. The end-
pieces serve as interfaces between the sample and upstream and downstream flow lines.
During a two-phase steady-state flow measurement, two fluids are injected into and
produced from the sample until steady-state is attained, that is, until saturation and

 SCA2016-001 2/12

pressure drop across the length of the sample stabilize. The difference in phase pressures
is close to zero at the outlet of the sample because of the capillary end-effect
phenomenon (CEE). The difference between phase pressures is non-zero at the core inlet.
From independent measures of injected phase pressures, one could conceptually
determine capillary pressure as the difference between the non-wetting and wetting phase
pressures. This is typically not practical because of phase mixing. This work addresses
the mixing issue, enabling measures of phase pressures at the sample inlet face.
In a steady-state test, ability to measure differences in injection phase pressures at the
sample inlet and to correct such measurements for viscous effects enables interpretation
of capillary pressure. Thus, this work is an effort towards measuring relative permeability
and capillary pressure curves on the same core plug (or first of a series of stacked plugs)
simultaneously. Wettability can be calculated using established methods such as the
Amott and the USBM (US Bureau of Mines) wettability index methods if both imbibition
and secondary drainage capillary pressure cycles are available. Combining relative
permeability, capillary pressure and wettability measurements into one test can yield a
significant reduction in experimental time compared to measuring each separately.
Much of the previous work (for example, Longeron et al., 1995; Richardson et al. 1952;
Jennings et al., 1988, Virnovsky et al., 1995a, Virnovsky et al., 1995b) relates to
independent phase pressure measurements at the sample inlet using hydrophobic (oil-wet)
and hydrophilic (water-wet) porous disks or membranes. For such approaches, properties
of the disk or membrane allow pressure communication with one phase while the other
phase is excluded. Drawbacks include the difficulty of initiating a reservoir-condition live
fluid test without exceeding excluded-phase entry pressures of the porous disks or
membranes, and uncertainty about whether or not the disk or membrane will perform as
desired throughout the test. This work seeks to isolate the injection phases physically for
phase pressure measurement rather than isolation via disks and membranes.
Authors including Richardson et al. (1952) and Gupta and Maloney (2015) suggested that
capillary pressure may be quantified as the difference between non-wetting and wetting
fluid phase pressures at the inlet. However, they did not account for the need to correct
the viscous flow contribution to the pressure difference, which this work addresses. They
also did not describe a practical inlet end-piece design to accomplish phase pressure
measurements. Richardson et al. (1952) stated that the difference between non-wetting
and wetting phase pressures at any point within porous media equals the capillary
pressure corresponding to the saturation at that point. They demonstrated the concept by
cementing gas (non-wetting) pressure probes to the rubber sleeve and oil (wetting phase)
pressure probes made of ceramic porous media to core walls. Their experiments showed
that the pressure difference between the non-wetting and wetting fluid phases inside the
core is constant away from the outlet end and equals capillary pressure. Cementing
probes on a sample is not a preferred approach because the practice might damage the
sample or alter its wettability. Further, cementing probes for each test could be time
intensive and susceptible to leaks.

 SCA2016-001 3/12

Kokkedee (1994) and Pini (2013) proposed that capillary pressure is equal to pressure
drop across the core at low rates. No special end-piece is required in this technique.
However, this technique assumes that viscous forces are small compared to capillary
forces, which may not be true in many test conditions. In this work, a method to correct
viscous forces from inlet phase pressure is used to estimate capillary pressure.

APPARATUS DESIGN
Design elements of the proof of concept apparatus are shown in Diagram 1. In this
design, end screens at the upstream face of the sample ordinarily used to promote mixing
of injection phases are omitted. The inlet face of the sample is directly in contact with the
inlet end-piece. To accurately measure difference in phase pressures at the inlet, two
differential pressure transducers (high and low pressure ranges) are used in parallel,
which can be engaged or disengaged depending upon the magnitude of the pressure
difference. Pressures are also measured with Quartz absolute pressure transducers to
provide redundancy. Pressure taps are placed as close as possible to the core holder inlet
to minimize pressure drops from flow in the tubing and are placed at the same height to
avoid gravity head differences. It is preferred to have check valves upstream of the inlet
side pressure transducers to prevent back flow, which could result in phase mixing. As in
a conventional flow system design, a differential pressure transducer or pair of Quartz
absolute pressure transducers is used to measure pressure drop across the length of the
sample. Pore pressure is maintained via closed-loop flow or by using a back pressure
regulator (BPR). Overburden pressure is supplied by a pump.

INLET END-PIECE DESIGNS

Designing an inlet end-piece that is robust and prevents injection phase mixing at rock
and end-piece junction is a challenging problem. Conventional inlet end-pieces have
patterns that promote distribution of fluid phases. Examples of several conventional end-
pieces are shown in Figure 1. Because test samples are porous, it is challenging to obtain
a good seal between a metallic end-piece and a rock sample. Small scale irregularities on
the rock surface and the smooth metal pattern on inlet end-pieces seal imperfectly,
resulting in mixing of injected fluids and equilibration of their pressures. The challenge
of obtaining a good seal and phase isolation exists even for a metal-based inlet end-piece
with flow distribution patterns that do not intersect (Figure 1b and Figure 1c). To
demonstrate this challenge, steady-state coreflood tests were performed on Cordova
Cream limestone (7-9 mD) using helium gas and brine (20000 ppm) at room temperature,
1500 psi pore pressure, and 2400 psi net confining stress. Metal inlet end-pieces like
those of Figure 1 were used. Capillary pressure was measured on a companion plug using
the centrifuge method with the same net confining stress and temperature conditions. The
modified flow apparatus, as discussed above, was used to perform this experiment. The
steady-state coreflood was performed with multiple gas-water fractional flows. Pressure
and saturation values were recorded at steady-state for each fractional flow. Figure 2
compares difference of inlet phase pressures for each fractional flow with centrifuge
capillary pressure data. Clearly, no inlet phase pressure difference (∆P) at steady-state
conditions was measured when using a metal-based inlet end-piece, indicating phase

 SCA2016-001 4/12

mixing at the inlet face of the sample. Similar results were observed from a repeat test
using a screen between the inlet end-piece and core.
New end-piece designs were tested with the aim of finding a configuration that prevents
phase mixing at the core inlet. New designs used machined metal end-pieces and
elastomer seals. The metal is used to provide: 1) a desired pattern for fluid distribution,
and 2) a rigid base for an elastomer seal that maintains structural integrity at high
pressures. The elastomer in the design is used to seal between the inlet end-piece and
sample surface to prevent phase mixing. The elastomer and metal/alloy needs to
withstand test conditions. Examples of a few end-piece designs are shown in Figure 3.
Although the combination of metal and elastomer can prevent phase mixing, some
designs perform better than others. For example, designs with small flow apertures like
the O-ring design (Figure 3a) provide excellent sealing between the inlet end-piece and
core face, but at the cost of significant additional viscous pressure contribution to the
inlet phase pressure. This additional viscous pressure contribution can exceed capillary
pressures. It comes from an enlarged region of high injection fluid saturation inside the
core near the inlet. Figure 4 shows Cordova Cream limestone results of pressure
difference of inlet phases from tests with the O-ring design and metal-based designs. In
this test, steady-state was attained with multiple total flow rates for two helium fraction
flows (0.5 and 0.8). The pressure difference of helium and brine were non-zero, clearly
indicating that the O-ring end-piece provided a good seal. However, helium-brine
pressure differences at the inlet were of large negative magnitude compared to centrifuge
capillary pressures. This resulted from the additional viscous pressure contribution due to
the O-ring design’s small inlet aperture. With the O-ring design, phase saturations
expanded in a hemispherical pattern inside the core at phase inlet ports. Since brine
viscosity is almost two orders of magnitude higher than helium viscosity, the viscous
pressure drop near the brine end-piece port is significant compared to that of the helium
port. For this reason, the difference in helium-brine inlet pressures makes it seem that
capillary pressure is negative in Figure 4. Typical measurement techniques for capillary
pressure are performed with no flow and negligible viscous pressure gradients. In this
test, the range of measured phase pressure differences match closely with Darcy equation
calculations for hemispherical flow in porous media. To remedy this problem, it appeared
desirable to have an elastomer-based inlet end-piece with wider flow aperture to reduce
the viscous pressure contribution in measured inlet phase pressures.
The use of an elastomer seal that withstands test fluids and conditions is important. Some
elastomers may perform well at room conditions, but may lose structural integrity in the
presence of hydrocarbons at high pressure and temperature. An example is shown in
Figure 5 in which a spiral elastomer pattern made using an incompatible material was
destroyed when exposed to live crude oil and brine at high temperature and pressure.
Inlet end-piece patterns with metal and elastomer (e.g., Half-moons and Spiral with
gasket in Figure 3) tend to perform better than those with only elastomer between the
end-piece and sample (e.g., Elastomer spiral). Elastomers are more compressible

 SCA2016-001 5/12

compared to metal. In a 100% elastomer based design, if grooves are not sufficiently
deep or wide, there is a risk of pattern distortion under high pressure. During another test
with Cordova Cream limestone outcrop at 1500 psi pore pressure and 2400 psi net
confining stress, an Elastomer spiral pattern was able to isolate phases at the inlet for the
first 4000 minutes (Figure 6), but later failed as evidenced by the reduction in inlet phase
pressures to zero as a result of phase mixing. Figure 7 shows post-test pictures of the core
and end-piece. The salt residue from brine spreading over most of the inlet end-piece face
(Figure 7a) and the brine streak connecting positions of injection ports for both phases on
the rock face (Figure 7b) provide clear evidence that fluids mixed at the core face and did
not remain isolated. The risk of having the elastomer lose structural integrity under high
pressure can be reduced if the end-piece is made of both elastomer and metal.
Based on the above discussion, a good inlet end-piece design contains: 1) a wider
aperture or surface area for injection phases to minimize the viscous pressure contribution
in measured inlet phase pressures, 2) compatible elastomer material, and 3) face design
consisting of both metal and elastomer, or 100% elastomer with sufficient groove width
and depth. Examples of such end-pieces are the Spiral with gasket (Figure 3c) and Half-
moons (Figure 3d).
From a similar experiment on Cordova Cream limestone outcrop, the Half-moon inlet
end-piece successfully isolated both brine and helium phases. Figure 8 shows the
difference of phase pressures at steady-state for multiple fractional flows and multiple
rates at each fractional flow. Clearly, phase pressure differences at the inlet are non-zero,
indicating successful isolation of injection phases at the face of the inlet end-piece.
Further, the differences between phase pressures at the inlet are positive, indicating lower
viscous contributions to the measurements because of better phase spreading at the core
plug face (or wide injection cross section) compared to results from an end-piece with
small injection aperture, such as the O-ring design. Though viscous contributions with the
Half-Moon design are relatively low, they are not zero, which reflects flow rate
dependence in the difference between inlet phase pressures. Because of viscous gradient
effects, the difference in inlet phase pressures is lower than the static capillary pressure
measured by the centrifuge method in Figure 8. Thus, the viscous contribution must be
subtracted from the difference in inlet phase pressures to achieve a measure of capillary
pressure from a steady-state coreflood test.

VISCOUS PRESSURE CORRECTION

The difference between non-wetting and wetting phase pressures at the inlet equals Pc
only at a static condition (zero flow rate). However, in a coreflood test (both steady-state
and unsteady-state), one or both phases are continuously injected, and the phase pressures
at the inlet result from both capillary and viscous forces. Hence, a method is required to
estimate and correct the viscous pressure contribution from the difference of phase
pressures at the inlet.
In a steady-state coreflood test, corrections for viscous pressure contributions can be
determined by performing tests at multiple flow rates for each fractional flow. Steady-

 SCA2016-001 6/12

state is achieved at each flow rate. For a given fractional flow, capillary pressure is equal
to the intercept of the trend line from a plot of steady-state phase pressure difference at
the inlet versus total flow rate (Figure 9). Close to the inlet, fluid saturation does not
change significantly with flow rates. Hence, phase pressure difference at the inlet changes
linearly with total flow rate for a typical steady-state fractional flow condition, and the
intercept of the linear trend equals capillary pressure. The capillary pressure measured
using the above method corresponds to the capillary end-effect corrected fluid saturation,
which can be estimated using the Intercept Method (Gupta and Maloney, 2015) or
through in-situ saturation monitoring. In tests with sufficiently high pressure drop across
the core, the correction might be small and within experimental accuracy, and the average
saturation across the core can be used.

APPLICATION EXAMPLE AT RESERVOIR CONDITIONS

A water-oil steady-state coreflood test was performed using one of the new inlet end-
pieces to demonstrate the concept of obtaining capillary pressure from the difference of
phase pressures (after viscous pressure correction) at the inlet. This test was performed
on native state core plugs stacked in series to make a composite of 25 cm length. The test
was performed at reservoir temperature and reservoir pore and overburden pressures
using live fluids. Viscosities of oil and brine were similar. Pressure measurement devices
were configured as in Diagram 1. The “Half-Moons” inlet end-piece was used (Figure
3d). The test was an imbibition steady-state test, beginning with a measurement of oil
permeability at irreducible water saturation and continuing with oil and water co-
injection in steps of decreasing oil fractional flow (Fo). For each Fo, after achieving
steady-state conditions, total flow rate was increased several times. With the direction of
total flow rate increasing, unwanted hysteresis is avoided. When Fo is decreased for the
next set of steady-state measurements, flow rate can be reduced concurrently with the
fractional flow change. Because the change in oil fractional flow coincides with an
increase in brine saturation, the reduction of total flow rate between consecutive
fractional flows avoids or reduces potential for inadvertently introducing hysteresis
because of saturation reversals.
Figure 10 shows phase pressure difference (oil minus brine phase pressure) at the inlet
and total flow rate versus time. As discussed above, reductions in total flow rate on
Figure 10 coincide with changes in fraction flow. In Figure 10, the difference in phase
pressures at the inlet is non-zero and ranges from -10 to +10 psi. The “Half-Moons” end-
piece was able to clearly prevent phase mixing at the inlet face during this test.
In theory, the capillary pressure contribution to the difference between measured phase
pressures is insensitive to total flow rate. In Figure 10, the phase pressure difference at
the inlet changes with flow rate at each fraction flow. This flow rate dependence is the
result of viscous rather than capillary phenomena. Figure 11 plots oil phase pressure
minus brine phase pressure at the inlet versus total flow rate from steady-state
measurements with oil fractional flows of 0.985, 0.9 and 0.3. All three plots show linear
trends, as do plots for the other fractional flows that are not shown. Intercepts equal
capillary pressure, while slopes are influenced by viscous effects. Slopes are positive for

 SCA2016-001 7/12

oil fractional flows above 0.5 (because oil rates are greater than brine rates) and negative
for fractional flows below 0.5 (because brine rates are greater than oil rates).

Constructing the capillary pressure curve for the rock at the inlet consists of plotting the
intercepts (from the phase pressure difference versus rate plots) versus the corresponding
brine saturations. Brine saturation was calculated by applying the Intercept method
(Gupta and Maloney, 2015), which corrects the steady-state data for capillary end-effects.
Since pressure drops in this test were high, capillary end-effect corrections to brine
saturations were small. Figure 12 shows the resulting capillary pressure plot for this test.
Saturations at the first and last points represent residual brine and residual oil saturations
for this test. In theory, capillary pressure curves asymptote at residual saturations at both
ends, but values of 11 psi and -15 psi were used to denote the capillary pressures at these
residual saturations. This plot closely matches the centrifuge capillary pressure curves
measured on preserved companion plugs of the same rock type (Figure 12), particularly
with core sample C1. Figure 13 shows pore throat size distributions for the centrifuge
core plugs (C1 to C4) and inlet-plug (S1) of the steady-state test. Clearly, plug C1 pore
throat size and imbibition centrifuge capillary pressure data is closest to core plug S1
used in the steady-state test. The kr curve obtained in this test closely matched with the
curve obtained on samples with similar geology.

Overall, the new method is easy to implement and only requires minor modifications to a
steady-state apparatus along with using a modified inlet end-piece. Even if an end-piece
fails to prevent phase mixing during a test, quality steady-state relative permeability data
is still obtained. However, this method has certain challenges. For example, the inlet
end-piece design should be robust and needs to prevent phase mixing throughout the test,
the test duration is slightly increased due to multiple flow rates performed for each
fractional flow, and viscous pressure correction needs to be added to the test workflow.
The application for a gas-liquid system is relatively more challenging compared to a
liquid-liquid system because of low gas viscosity. A gas is more susceptible to leaks and
phase mixing and generates relatively high viscosity pressure correction compared to a
liquid.

CONCLUSION
Modified inlet end-pieces described in this work successfully isolated injected phases in a
coreflood test, allowing measurement of inlet phase pressures. The inlet phase pressures
at steady-state conditions were used to estimate capillary pressure of the core after
applying viscous pressure corrections. If both imbibition and secondary drainage cycles
are performed during a steady-state test, then capillary pressures as well as relative
permeabilities can also be measured. Wettability of the core can be inferred from
imbibition and secondary drainage capillary pressure curves using either Amott or USBM
wettability index methods. Thus, using the described inlet end-piece designs and test
methodology, relative permeability, capillary pressure and wettability of a core can be
measured simultaneously in a single steady-state test, which may result in significant
time and cost savings.

 SCA2016-001 8/12

REFERENCES
1. Richardson, J. G., Kerver, J. K., Hafford, J. A. and Osoba, J. S. (1952, August 1).
Laboratory Determination of Relative Permeability. Society of Petroleum Engineers.
doi:10.2118/952187-G
2. Longeron, D., Hammervold, W. L. and Skjaeveland, S. M. (1995, January 1). Water-
Oil Capillary Pressure and Wettability Measurements Using Micropore Membrane
Technique. Society of Petroleum Engineers. doi:10.2118/30006-MS
3. Virnovsky, G.A., Guo, Y. and Skjaeveland, S.M. (1995a, May 15). Relative
Permeability and Capillary Pressure Concurrently Determined from Steady - State
Flow Experiments. 8th. European IOR - Symposium in Vienna, Austria
4. Virnovsky, G.A., Guo, Y., Skjaeveland, S.M. and Ingsoy, P (1995b, October).
Steady-State Relative Permeability Measurements and Interpretation with Account
for Capillary Effects. Society of Core Analysts Conference in San Francisco, USA
5. Gupta, R., and Maloney, D. R. (2015, December 1). Intercept Method--A Novel
Technique To Correct Steady-State Relative Permeability Data for Capillary End
Effects. Society of Petroleum Engineers. doi:10.2118/171797-PA
6. Kokkedee, J. A. (1994, January 1). Simultaneous Determination of Capillary
Pressure and Relative Permeability of a Displaced Phase. Society of Petroleum
Engineers. doi:10.2118/28827-MS
7. Pini, R., and Benson, S.M.. 2013. "Simultaneous Determination of Capillary
Pressure and Relative Permeability Curves from Core-Flooding Experiments with
Various Fluid Pairs." Water Resources Research 49 (6): 3516–30. doi:
10.1002/wrcr.20274

Diagram 1: Schematic of modified core holder assembly

 SCA2016-001 9/12

(a) (b) (c)

Figure 1: Examples of conventional inlet end-piece designs: (a) Cross, (b) Spiral, and (c) Modified spiral

Figure 2: Difference of inlet phase pressure compared to measured capillary pressure for Cross, Spiral and Modified
spiral inlet end-pieces

(a) (b) (c) (d)

Figure 3: Examples of new elastomer based inlet end-piece designs, (a) O-ring, (b) Elastomer spiral, (c), Spiral with
gasket, and (d) Half-moons

Viscous ∆P range from hemispherical

Darcy’s flow

Figure 4: Difference of inlet phase pressure compared to measured capillary pressure for Cross, Spiral, Modified
spiral and O-ring inlet end-pieces

 SCA2016-001 10/12

Figure 5: Spiral with gasket end-piece after exposure to live crude oil at high pressure and temperature.
Incompatible material was used in this test.

Figure 6: Plot of difference in inlet phase pressures with time for helium-brine steady-state test with Elastomer
spiral inlet end-piece

Brine salt Brine imprint

covers most indicating
of the face phase mixing

Figure 7: Pictures taken after the test for, (a) face of Elastomer spiral inlet end-piece, and (b) face of the core at the
inlet side

Figure 8: Difference of inlet phase pressures from Half-Moons end-piece compared to measured capillary pressure

 SCA2016-001 11/12

Figure 9: Plot of steady-state phase pressure difference at inlet with total flow rate for a given fractional flow. The
intercept of the trend line of this plot equals static capillary pressure.

Figure 10: Difference in phase pressure at inlet and total flow rate with time at different oil fractional flow (Fo)

Figure 11: Plot of inlet phase pressure difference at steady-state condition with total flow rate for oil fractional
flows of 0.985, 0.95 and 0.3. All three plots have linear trends. Intercepts equal capillary pressures.

 SCA2016-001 12/12

Figure 12: Capillary pressure curve obtained using the new method and compared with centrifuge capillary
pressure measured on preserved plugs (C1 -C4) of the same rock type

Figure 13: Pore throat size distribution for the inlet core plug used in the steady-state test and the four core plugs
for same rock type used in the centrifuge capillary pressure test

 SCA2016-023 1/12

SCAL FOR GAS RESERVOIRS: A CONTRIBUTION FOR

BETTER EXPERIMENTS
Arjen Cense1, Jules Reed2, Patrick Egermann3
1
A/S Norske Shell, 2LR Senergy, 3Storengy (ENGIE)

This paper was prepared for presentation at the International Symposium of the Society of Core
Analysts held in Snowmass, Colorado, USA, 21-26 August 2016

ABSTRACT
Special core analysis for gas fields requires a different approach than for oil fields. Gas
fields are mostly produced via pressure depletion. Saturation changes are not large during
this process, except if water is flooding (parts of) the reservoir. Gas reservoirs are mostly
water wet, which is, from a SCAL point of view and from a reservoir simulation point of
view, a significant simplification compared to oil fields, where wettability is often a big
uncertainty and has significant impact on the shape of the (imbibition) capillary pressure
and relative permeability curves. However, although the required SCAL input for
reservoir simulation may seem to be relatively simple, the design and do-ability of the
SCAL experiments in the lab is challenging, due to compressibility and solubility effects
of gas. In this paper, we discuss what laboratory experiments are appropriate for gas
fields and how they should be performed to obtain maximum value. We discuss what the
uncertainties in these experiments are and how uncertainties can be reduced.

INTRODUCTION
With the current low oil prices and the focus to reduce cost, investments in gas projects
are becoming more attractive. The price of gas has not dropped nearly as much as that of
oil, which makes gas producing fields economically more stable. In addition, at the last
climate change summit in Paris, it was decided by leaders of 195 nations that they will
cut their carbon emissions. Since gas emits lower carbon emissions than coal and crude,
gas producing fields are potential candidates to decrease the pace of global warming until
we can rely on renewable energy sources.

When a natural gas field is discovered, the initial distribution of gas can often be inferred
from the primary drainage capillary pressure curves, from which the saturation height
function, i.e. gas saturation as a function of height-above-free-water-level (HAFWL), can
be established. If the water level has moved pre-production, because of a (temporary)
breach in the seal or because of a hydrodynamic (flowing) aquifer, we need to use the
imbibition capillary pressure curves to describe those parts of the reservoir where water
has displaced gas, and where gas has become residual or trapped (Figure 1).

Most natural gas fields are produced by depletion, meaning that wells produce gas by
lowering the reservoir pore pressure, until production becomes uneconomic. Gas
reservoirs are generally water wet, as the rock has not been in contact with wettability
altering hydrocarbon components such as resins and asphaltenes. The processes at play
 SCA2016-023 2/12

during reservoir production are simple: saturations hardly change and gas is flowing at
initial (or connate) water saturation. The gas relative permeability is determined as the
endpoint gas relative permeability (krg(Swi or Swc) - green point in Figure 2). The water is
still immobile, hence water relative permeability is zero.

In cases where a (strong) aquifer is present, the aquifer will invade during pressure
depletion (depicted by orange arrows in Figure 1). The water from the aquifer then
displaces the gas to its endpoint (yellow point in Figure 2), while the gas becomes
trapped at residual gas saturation (Sgr). We generally see that pressures in gas reservoir
models are mostly sensitive to krg(Swi). Ultimate recovery appears less sensitive to
krg(Swi), but is more a function of Sgr and the endpoint water relative permeability,
krw(Sgr). The shape of the relative permeability curves (e.g. gas and water Corey) are
considered to be less important.

Figure 1: Schematic picture of gas reservoir Figure 2: Relative permeability as function of

with one producing well water saturation

PITFALLS
Capillary Pressure
Generally speaking, the capillary pressure (Pc) curves are important to acquire because
they have implications on both the volume of hydrocarbon in place and the sweep
efficiency in the case of an active aquifer.

Gas In Place (GIP): the drainage Pc curves are usually the most relevant to evaluate the
GIP, accounting for the transition zone especially for low-medium permeability range
reservoirs. Nevertheless, for some reservoirs, the current Free Water Level (FWL)
identified from the gas-brine pressure gradients can be different from the one prevailing
several millions years before which is often referred to as paleo contact. In the situation
where the paleo contact was shallower than the current contact, the system is still under
gas filling process and therefore the drainage mode is still the relevant one (gas saturation
increase). In the opposite situation, a paleo contact located deeper than the current contact
means that the whole system has experienced a major imbibition process. It is therefore
important to keep in mind that drainage Pc curves are not always the relevant curves for
 SCA2016-023 3/12

GIP evaluation. There are two types of information that can provide indications about the
possible paleo contact:
● non-zero gas saturation below the FWL, from pressure gradient measurements,
indicating gas saturation values typically in the trapped gas saturation range (20-
40%).
● gas saturation in the transition zone not properly reproducible by a drainage
J-function.

An illustration of such paleo contact data is provided in Figure 3. In this case, a strong
deviation from the drainage J-function trend was clearly observed in a depth window up
to 12 ft above the FWL (left plot). Above this depth, the deviation between imbibition
and drainage curves is narrower, in the asymptotic part (right plot) of the Pc curve. In this
case, the transition zone impact was of limited extent because of favorable permeability,
but it can lead to much more pronounced deviation for lower permeability reservoirs. For
flat and extended reservoirs, the volume impact of using a drainage instead of an
imbibition curve can be huge, because of this saturation difference in the transition zone.
Furthermore, the amount of gas trapped in the residual gas zone, below the present day
free water level, will still be influenced by depletion of reservoir pressure and hence, will
provide energy to the producing reservoir, via either the brine phase or the gas phase. The
residual gas bubbles will expand upon depletion and once a connected path flow is
established, the gas will reconnect with the ‘free’ gas in the reservoir.

Figure 3: Example of paleo contact detected from a deviation of the saturation behavior in the
transition zone

Sweep efficiency: because gas-water systems are always strongly water-wet and
displacement is piston-like, the shape of Pc imbibition curve has little impact on the
waterflood behavior, in terms of breakthrough time and pressure, except for the Sgr
endpoint value. At (very) low injection rates, the core may spontaneously imbibe water,
leading to negative pressure drops across the core.

In both cases, a preliminary estimate of the imbibition Pc curve can be obtained by

rescaling the drainage curve between Swi and Sgr as soon as the Sgr value is consistent (see
next section). It is possible to infer the drainage curve by deriving the J-functions through
 SCA2016-023 4/12

various standard techniques: HPMI (High Pressure Mercury Injection), centrifuge or

porous plate.

In all the cases, the Pc curves to use for reservoir conditions purposes must account for
both stress and interfacial tension using the Leverett formalism. Specific additional
experiments are usually required like permeability-porosity as a function of stress and
IFT data using either direct values from the pendant - ascending drop technique or
estimates from PVT calculation, if the full gas and oil compositions are known. Among
the 3 Pc techniques, porous plate is the only one to obtain direct relevant drainage Pc
curves, if the experiments are conducted under full reservoir conditions. Usually the 3
approaches compare reasonably well for conventional gas-liquid reservoir systems:
permeability > several mD (Sabatier, 1994) and the HPMI - centrifuge corrected curves
can often be considered as a fair estimate whilst awaiting the porous plate results, which
can take several months.

Residual Gas Saturation: Sgr is one of the most important parameters to acquire for an
imbibition liquid-gas scenario, because it controls both the final saturation state and the
dynamic behavior, since the displacement is piston-like (hence Sgr affects the
breakthrough time). Representative Sgr values may be acquired from coreflood
experiments conducted under stress, with some pore pressure to limit gas compressibility
effects (see section on relative permeability endpoints). Sgr can be calculated by simple
material balance (volumetric or gravimetric) and/or direct measurements using in-situ
saturation monitoring (ISSM), which gives the added bonus of observing saturation as a
function of sample length; although, uncertainties can be as high as several saturation
units (s.u.), especially at reservoir conditions (Cense et al., 2014). In general, Sgr values
range between 20 and 40 s.u., where such inaccuracy can lead to errors typically in the
range of 3% original GIP. It is therefore of primary importance to use several approaches
for QC evaluations.

Firstly, include a compressibility test in the protocol after imbibition, in order to directly
evaluate the remaining gas volume in the sample. Close the sample outlet and inject the
imbibition test liquid until a predetermined pressure is achieved. Since pressure increase
is linked directly to gas compression, induced by the liquid volume increase, it enables
determination of the gas volume in place using the perfect or real gas law, dependent
upon the fluid system.

With Pini and Ptest respectively, the initial pressure and stabilized pressure after injection,
the initial gas volume (prior to compressibility test), the volume of liquid
injected and PV the pore volume. This simple technique can give an additional and
independent evaluation of Sgr very rapidly after the experiment.
 SCA2016-023 5/12

Secondly, compare the results with published compilations of SgrM (maximum Sgr
obtained with Swi equal to 0) acquired from various samples (Suzanne, 2003; Pentland,
2010). As Sgr must be compared with SgrM, the Land equation (1971) can be used for the
conversion, using default values of the Land constant according to the rock type (Irwin
and Baticky, 1997).

Thirdly, conduct a large number of ambient conditions SgrM experiments in parallel to the
few imbibition coreflood experiments planned. These experiments are simple, fast and
relatively cheap to conduct. Clean samples are dried at 60°C and weighed. Immerse
individual samples in a beaker of brine for roughly one hour for standard permeability
(longer times may begin to introduce artefacts linked to gas dissolution, (Suzanne,
2003)). Alternatively, monitor immersed weight as a function of time, to the point of
inflexion. After imbibition, take the samples from the beaker, remove superficial water
and weigh. SgrM is calculated as the weight difference from dry state.

Figure 4: Probability curve of SgrM obtained from 73 measurements (left) and derived Sgr values
using Land (1971)

An example of this approach is provided in Figure 4 (above), where an SgrM cumulative

probability curve was obtained from 73 measurements on the same facies. It enabled
definition of P50 and the associated variability. Using the Land equation (Land, 1971)
and a representative Swi value (20% considered in this example), it was possible to derive
the range of expected Sgr values and to compare them with those obtained during the
coreflood experiments. Another benefit of this approach is that the cumulative probability
curve can be used to define the range of variation of the Sgr parameter, if the reservoir
model is history matched through an automatic process.

The Brooks & Corey formalism (1966) is very attractive in the case of gas-liquid
systems, since there is no ambiguity about the wettability. Drainage Pc data (mercury
injection, centrifuge or porous plate) are first matched using the parametric law:
 SCA2016-023 6/12

where Pd is the threshold pressure, λ rules the shape of the curve and represents the throat
size distribution, Sr is the residual saturation. Once matched, the drainage gas-liquid
relative permeability curves can be deduced in a straightforward manner using the
following formulae:

Although not reported explicitly in the literature, the Brooks and Corey approach (1966)
can be easily applied to analyze imbibition data just by coupling the non-wetting phase
trapping formula introduced by Land (1971) with a Carlson type hysteresis model
(Carlson, 1981). For KrgI (imbibition), we use the above equation for drainage (D) in
parallel with KrgI(Sg) = KrgD(Sgf), Sgf being the free gas saturation given from Land
! ! !
equation through the relation 𝑆!" = ! 𝑆! − 𝑆!" + 𝑆! − 𝑆!" − ! 𝑆! − 𝑆!"
The Land formula must be calibrated to obtain the C constant representative of the facies
! !
under concern, where: !
−! =𝐶
!" !!

Once calibrated (C, Swi and λ), it is possible to assess the imbibition relative permeability
curves and especially the endpoints. Figure 5 shows an example of qualitative QC using
this approach on the endpoints of gas-brine relative permeability data. The blue and red
curves provide the range of the expected Krw@Swi and Krg@Sgr values, respectively. In
this case, several outliers can be easily detected for further analysis of the raw data and
interpretation.

Another practical approach is to compare relative permeability data from independent

techniques. When Pc data are obtained from centrifuge, for example, it is recommended
to spend some extra time to interpret the relative permeability curve of the displaced fluid
in order to compare it with the one obtained from displacement technique (SS or USS).
Where good agreement is observed, it adds confidence to the shape of the relative
permeability curves, especially at late saturation, as depicted by the example provided in
Figure 5 (right).
 SCA2016-023 7/12

Figure 5: Quality control of endpoint relative permeability using Brooks and Corey type approach
(1966) (left) – comparison between USS and centrifuge data for QC purpose (right)

Low Rate Waterflooding – Unsteady State

Bull et al., 2011 presented data indicating that Sgr was a function of injection rate and
recommended that residual gas measurements be performed at elevated or reservoir
pressure to minimize gas compression and diffusion effects. However, many laboratories
continue to perform analyses with little consideration of the implications of injection rate,
compressibility and disequilibria (i.e. gas and water are not in full equilibrium due to
pressure differentials required for flooding). Pore pressure for these tests in commercial
labs, is often between 200-300 psi (15-20 bar) because of the pressure limitations of the
graduated glass separators used to measure volumetric changes of the fluids. ISSM is
always recommended as a secondary, verifying measurement method.

A number of unsteady state waterflood displacements to Sgr were reviewed. Pore

pressures were 200 psi for all samples. Water injection rates varied from 0.5-4 ml/min,
with little correlation to core properties. Figure 6 is a plot of permeability (normalized
for area and length) against flow rates employed by a single lab during two projects from
the same field, indicating the diversity and ill-considered rates employed. Figure 7
depicts a plot of Sgr versus water differential pressure (DP) after breakthrough for these
samples. There is an obvious decline in residual gas saturation once DP exceeded 10 psi
(5% of the applied pore pressure). This data confirms recommendations to limit
differential pressure as a function of applied pore pressure (McPhee et al., 2015), though
the book states a more stringent limit of 2% pore pressure.

Low rate flooding experiments are performed to obtain Sgr values for cores initialized at
representative Swi (by porous plate or by centrifuge). When performing a waterflood on a
gas filled core at Swi, the gas volume produced equals the water volume injected until
breakthrough. At breakthrough, gas saturation is observed to be residual throughout the
core, and thus, no further gas production is expected after breakthrough.
100000 0.35

0.3
Residual (trapped) Gas Saturation

Reducing Sgr
10000
for DP > 10 psi
0.25 (3-5 % PP)

1000
0.2
K·A/L

0.15
100

0.1

10
0.05

1 0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 0.01 0.1 1 10 100 1000
USS Waterflood flow rate, ml/min Differential Pressure after breakthrough, psi

Figure 6: Permeability (normalized for Area & Figure 7: Sgr versus DP measured during post-
Length) versus Lab applied flow rates. breakthrough waterflooding
 SCA2016-023 8/12

However, in some experiments, gas production continued after breakthrough. Plotting gas
production as a function of brine pore volumes injected (PVI), an average (4
experiments) 2.9 ml additional gas was produced after approximately 30 PVI (ca. 400 ml)
of brine after breakthrough (see Figure 8 (left)). Gas production was confirmed by ISSM.
At breakthrough, the profiles were flat, i.e. the water saturation in the core is constant.
After flooding multiple pore volumes, the water saturation has increased throughout the
core, but more at the inlet than at the outlet.

Once brine is injected in the core, the pressure increases from 20 bar (back pressure of
the whole system) to an injection pressure between 20.01 and 20.08 bar (differential
pressure of 0.01-0.08 bar). Due to the higher inlet pressure, more nitrogen can be
dissolved in the brine. The dissolved nitrogen will be evolved downstream of the core as
pressure drops towards 20 bar again. It will either be trapped as residual gas, or surplus
gas will be produced by viscous forces of the flowing brine.

In an attempt to quantifying this effect, an average pressure increase of 0.04 bar (4000
Pa) was assumed. The increased pressure leads to a concentration change of 4000 Pa /
(155·106 Pa / (mol/l) [Wilhelm et al. (1977)] = 2.58·10-5 mmol/l. The amount of nitrogen
lost during 400 ml of brine flushing through, at the inlet of the core is 1.03·10-5 mmol, at
room temperature and 20 bar pressure 1.3·10-5 ml of nitrogen. This volume loss is orders
of magnitude lower than the observed loss of 2.9 ml of nitrogen.

Another effect contributing to the gas stripping at the inlet happens when small amounts
of gas are dissolved in the by-passing brine, which is under saturated: the size of the
trapped gas bubbles is reduced, and as a consequence the local capillary pressure
increases as the capillary pressure is proportional to the local curvature of the gas
bubbles. This increases the local gas pressure and leads to gas dissolution, further
decreasing the bubble size.

Figure 8: (Left) Gas production as a function of pore volumes injected (PVI). Gas production was
measured using gamma ray (red line) & separator (dotted purple line). Pressure drop across the core
in grey. (Right) Water saturation as a function of distance from inlet. Flooding from left to right.
 SCA2016-023 9/12

In these experiments, the brine-nitrogen system was equilibrated for 24 hours by

bubbling the nitrogen through the brine and the pore pressure was 100 bar. In another
experiment, where the pore pressure was 5 bar and the equilibration time was 18 days, we
did not observe any production of gas after breakthrough, which leads to the conclusion
that 24 hours is not long enough to equilibrate nitrogen and brine from atmospheric
pressure to 20 bar.

The tail end production of gas is thus considered to be an experimental artifact. The
residual gas saturation representative for the reservoir is the residual gas saturation at
breakthrough. It is our recommendation that USS waterfloods be performed at equivalent
reservoir advancement rates (if the DP limits allow, else lower rates may be required), at
limited differential pressure (max 5% of pore pressure) and for a limited injection
volume, maximum 2 PVI.

Steady State Waterflooding

Steady state water displacing gas experiments are not generally recommended as a
method to acquire imbibition water-gas relative permeability data; unless one is aware of
the limited saturation range expected from these analyses and of the potential saturation
inaccuracies because of disequilibria effects, or unless one has access to appropriate
equipment and expertise, together with sufficient time and budget.

Figure 9 provides two plots from an imbibition water-gas steady state relative
permeability test. These plots depict the correlation between anomalous, non-uniform
saturation profiles (left) and differential pressures measured during each fractional flow
rate of the test (right), plotted as a function of pore volumes water injected (PVI). Water
saturation begins from Swi = 0.163 and increases to Sw = 0.505 during the first fractional
water flow (fw=0.003). This is an increase of 34.5 saturation units (s.u.) at a very low
fractional flow rate and represents one of the lowest viable fractional rates for most
commercial laboratories (equipment and time limitations). Saturation is relatively
uniform after this first fractional rate, and after the subsequent two fractions. However,
after fw=0.201, where DP increases above 10 psi, and above 5% pore pressure, a
significant decreasing saturation gradient is observed from injection to production face (0
to 1 fractional length). The gradient increases during the next two fractional flow rates
because of increased DP and greater water throughput. This results in inaccurate
saturation data for the last three fractions and a poorly defined relative permeability
saturation range, which defeats the purpose of the steady state method, since a major
objective is to define a wider relative permeability saturation range than obtained by the
unsteady state method.
 SCA2016-023 10/12

fw = 0.003 fw = 0.012 fw = 0.025 fw = 0.003 fw = 0.012 fw = 0.025

fw = 0.201 fw = 0.716 fw = 1 fw = 0.201 fw = 0.716 fw = 1
1 1000

0.9

0.8
Water Saturation, fraction

100

Differential Pressure, psi

0.7

0.6

0.5 10

0.4

0.3 Injection
DP = 10 psi
1 PP = 200-300 psi
0.2 DP/PP ratio = 3-5%
0.1

0 0.1
0 0.2 0.4 0.6 0.8 1 0 20 40 60 80 100 120 140
Fractional Length Water PVI

Figure 9: DP versus water PVI during an imbibition steady state waterflood (left). Resultant ISSM
profiles during this experiment, depicting disequilibria effects at the sample injection face.

The analyses are controlled strongly by imbibition capillary pressures, resulting in the
observed large encroachment of water saturation at very low fractional water flow rates.
The large saturation increase is unavoidable. Fractional flow rates require to be some
orders of magnitude lower, which would require either, ultra-low water flow rates (hence
extremely long experimental times), or higher total injection rates (requiring very high
gas injection rates, which risk extending into turbulent flow regimes).

Simulations of these steady state experiments were performed in an attempt to determine

potential experimental parameters that might provide reasonable results within the
limitations described earlier for standard commercial laboratory equipment, at ambient
temperature and pore pressures of approximately 300 psi (20 bar).

Corey functions of Ng=2, Nw=5, endpoint krw=0.1 and Sgr=0.25 were used together with
imbibition Pc data, modified from MICP curves, for a sample permeability range from 1-
100 mD. Simulation using an initial fractional flow rate of fw=0.0001, approximately a
magnitude lower than the SS experimental data above, resulted in the saturation profiles
given in Figure 10, for the 100 mD case. In this case, there was still a significant
saturation change (over 35 s.u.) during that first, low fw step. The first fw step was
estimated to require approximately 10 days to stabilize. A further, magnitude-lower fw
was estimated to require over 30 days to stabilize during the first fw. Such stabilization
times are uneconomical for both laboratory and client company, hence unfeasible. It was
also noted that saturation after subsequent fw steps was observed, at best, to cover a 25
s.u. range.

In addition to these undesirable saturation changes, the high throughput volumes required
during a steady state test lead to a greater potential for disequilibria effects. Initially, at
low fw, differential pressure is low because gas viscosity is the main flowing phase;
however, as fw increases, DP will increase and may exceed the DP limits described
above. This may be regulated by decreasing the total injection rate at each increasing fw.
 SCA2016-023 11/12

It is therefore recommended that, if employing gas-liquid SS testing (given the inherent

issues), DP should be minimized in this manner.

Figure 10: Example of simulated saturation profiles for SS water-gas relative permeability

CONCLUSIONS
In this paper we have discussed common pitfalls that may arise when conducting gas
water SCAL experiments. The difference between drainage and imbibition processes can
lead to an evaluation of too low gas saturation in place. This may be determined if there
is no match between log saturations and those inferred from drainage models or if there is
a deviation from J-function saturations. The porous plate technique is recommended as it
can measure the Pc curves at full reservoir conditions. However, mercury-air or
centrifuge measurements are much quicker to perform, and give reasonably reliable data.

It is recommended to use reservoir equivalent advancement rates (low rate) for corefloods
to Sgr. Fluid equilibration time, between gas and brine, is important prior to conducting
flooding experiments. It is recommended to limit differential pressures during low rate
corefloods, and limit injection to two pore volumes. It is recommended to compare and
quality control Sgr from (low rate) coreflood experiments, using ISSM, other volumetric
methods, and published and measured values for SgrM. Relative permeability endpoints
should be checked for outliers using an analytical approach that helps in determining
confidence intervals.

Water-gas imbibition steady state experiments are not recommended when using the
limited equipment of most commercial laboratories, unless one is aware of the limitations
and cost implications of performing these tests, or one has equipment and expertise
appropriate to acquiring these data.
 SCA2016-023 12/12

REFERENCES
1. Brooks, R. H. and Corey, A. T.: "Properties of Porous Media Affecting Fluid
Flow", J. Irrig. Drain. Div., (1966), 6, 61.
2. Land, C.: “Comparison of Calculated with Experimental Imbibition Relative
Permeability”, Society of Petroleum Engineers Journal, vol. 11, no. 4, pp. 419-425,
1971.
3. Pentland, C.H.: “Measurements of Non-Wetting Phase Trapping in Porous Media”,
PhD, Imperial College, November 2010.
4. Sabatier, L.: “Comparative study of drainage capillary pressure measurements
using different techniques and for different fluid systems”, SCA-9424.
5. Suzanne, K.: “Evaluation of the trapped gas saturation in water-wet sandstones –
Influence of the rock properties and the initial saturation state”, PhD, Ecole des
Mines de Paris, February 2003.
6. Honarpour, M. and Mahmood, S.M.: “Relative permeability measurements: an
overview”, Journal of Petroleum Technology (1988), SPE 18565.
7. Jerauld, G.R.: “Prudhoe Bay Gas/Oil Relative Permeability”, SPE Reservoir
Engineering, 1997
8. Cense, A.W., van der Linde, H.A., Brussee, N., Beljaars, J., Schwing, A.: “How
reliable is in situ saturation monitoring (ISSM) using X-ray?”, SCA-2014-09.
9. Wilhelm, E., Battino, R., Wilcock, R.J.: “Low-pressure solubility of gases in liquid
water.” J Chem. Rev., 77, 219–262, 1977.
10. Goodman, J.B., Krase, N.W.: “Solubility of Nitrogen in Water at High Pressures
and Temperatures” Industrial and Engineering Chemistry, 1931.
11. Bull, Ø., Bratteli, F., Ringen, J.K., Melhuus, K., Bye, A.L., Iversen, J.E.: “The
Quest for True Residual Oil Saturation – An Experimental Approach”, SCA201103
12. McPhee, C., Reed, J., Zubizarreta, I. “Core Analysis: A Best Practice Guide”,
chapter 10 about relative permeability, Developments in Petroleum Science,
Elsevier, 2015, Volume 64, Pages 519-653.
13. Irwin D.D., Baticky J.P. “The successive displacement process: oil recovery during
blowdown”, SPE 36719, November 1997, SPERE Journal.
14. Carlson F.M.: “Simulation of relative permeability hysteresis to the non-wetting
phase”, SPE 10157, ATCE, San Antonio, Texas, 4-7 October 1981.
 SCA2016-012 1/12

Gas-water steady-state relative permeability determination

with two approaches; experimental and digital rock analysis,
strengths and weaknesses
R. Farokhpoor1, E. Westphal1, N. Idowu1, P.E. Øren1, B. Fletcher2
1
FEI Oil and Gas, 2BG Group, Reading, UK

This paper was prepared for presentation at the International Symposium of the Society
of Core Analysts held in Snowmass, Colorado, USA, 21-26 August 2016

ABSTRACT
Digital Rock Analysis (DRA) or pore-scale imaging and modelling have developed
significantly in the last decade with the emphasis changing from phenomenological
research towards quantitative modelling. The evolution towards more predictive
modelling raises questions about the reliability of DRA derived properties compared to
experimentally measured data, especially for multiphase flow.
In this work, we compare gas-water relative permeability functions obtained from
experimental measurements with those derived from DRA. The steady state tests were
performed on single plugs with in-situ saturation monitoring. The experiments showed
strong non-uniformity in saturation profiles along the plugs and a numerical approach is
used to correct for this artefact.
3D X-ray micro-computed tomography (micro-CT) images of the three sandstone plugs
and representative sub-plugs were acquired along with Backscattered Scanning Electron
Microscopy (BSEM) images of 2D sections. 3D pore networks were extracted from the
sub-plug and process-based models (PBM). Gas-water primary drainage and water-gas
imbibition relative permeability curves were calculated on the different networks using a
pore network model simulator. The DRA derived relative permeability function displays
a large degree of similarity with the experimentally measured one. Comparison of
fractional flow curves between the experimental (before bump) and DRA computed
results shows a difference of 1.0 to 4.0 saturation units for recovery at breakthrough,
suggesting that the observed discrepancies in relative permeability have little implications
for practical applications.

INTRODUCTION
Accurate determination of relative permeability is essential in estimating producible
reserves and ultimate recovery in reservoir models. Laboratory measurements commonly
use flow rates higher than those characteristic of flooding processes in the reservoir to
overcome discontinuities in capillary pressure at the boundaries of the core plug.
Extensive studies have been performed to minimize this artefact during experiments
[1,2]. Numerous approaches have been presented for correcting experimental data using
numerical methods, semi-analytical approaches and/or inverse modeling [3-5].
Prediction of multiphase flow properties using DRA has evolved significantly in the last
decade and it is becoming a predictive tool in the oil industry. It is generally agreed that
 SCA2016-012 2/12

for mixed to oil-wet system, DRA cannot reliably predict multiphase flow properties
without detailed representation of contact angles, but for strongly water-wet systems
(gas-water systems in this work) DRA results should compare favorably with
experimental data. Some authors have compared DRA derived imbibition relative
permeability with experimental results for clean sandstones and sandpacks [6-11] but
further studies are required for more complicated samples.
In this study, we determined imbibition water-gas relative permeability on three
sandstone samples with two approaches. Flow simulations were performed on networks
which were extracted from a process-based model and on networks which were extracted
from the segmented micro-CT images of the sub-plugs. Comparisons of measured steady-
state relative permeability data with predicted results on the same samples show good
agreement and differences between the values are within an acceptable range.

Laboratory experimental data

Petrophysical properties of the three core plugs with short lithological descriptions are
given in Table 1.
Table 1: Petrophysical and lithological properties of the three core plugs
Plug Depth Porosity Permeability
Lithological properties
No. (m) (fraction) (mD)
Homogeneous, poorly consolidated, weakly sorted
Plug 1 3186.5 0.233 873
medium to coarse grained sand
Plug 2 4460.14 0.173 1.5 Homogeneous, well sorted fine sandstone
Plug 3 2684.84 0.289 2162 Poorly consolidated, coarse grained sand

Imbibition water-gas steady-state relative permeability experiments were performed on

the three core plugs in a core analysis laboratory (Figure 1).

Figure 1: Experimental imbibition water-gas steady-state relative permeability curves

 SCA2016-012 3/12

Saturation profiles were mapped during the core plug flooding experiments using an
AXRP-300, automated X-Ray relative permeameter. Figure 2 shows that there is a strong
non-uniformity in saturation profiles of plugs 1 and 2 while it is less pronounced for plug
3. To correct for this artefact, Sendra simulator was used. The core plug geometry and
petrophysical properties were used to build a simple model and to simulate a steady state
water-gas imbibition process. The main application is to determine relative permeability
from experimental data through an automated history matching of differential pressure
(Figure 3-left) and saturation profile. The relative permeability results based on LET [12]
correlation are plotted in Figure 3-right.

Figure 2: Water saturation profiles along the core plug during flooding experiments

Figure 3: Left) experimental differential pressure and Sendra history matched data for plug 1, Right)
comparison of experimental and Sendra relative permeability for plugs 1 and 2

Simulated Sendra results for plug 1 shows that residual gas saturation and water relative
permeability are lower compared to the experimental data. In plug 2, the Sendra history
matching predicted comparable residual gas saturation and water relative permeability to
the experimental data before the bump flood. The main disadvantage of correcting
artefact errors using a numerical simulation method is that the results are nonunique and
it depends on what correlation has been used.

Digital rock analysis

The workflow used for digital rock analysis can be summarized with the following steps:
1. Overview micro-CT scans of the received core plugs (19µm/voxel)
 SCA2016-012 4/12

2. Thin section preparation from a trim end of the core plug and BSEM imaging at
multiple scales
3. Image analysis to obtain input data for process-based modelling.
4. PBM construction (capturing relevant rock types).
5. Drilling of a sub-plug (2-5 mm in diameter and 1-2 cm long)
6. Scanning sub-plug by micro-CT at highest possible resolution (1-5 µm/voxel)
7. Processing and generation of rock models from the scanned images
8. Calculation of petrophysical properties (ø, Kabs).
9. Extraction of pore network representations from PBM and sub-plug 3D image.
10. Simulation of multiphase flow properties for gas-water system
The workflow is illustrated in Figure 4.

Figure 4: Schematic view of workflow

Imaging
Core plugs of 1.5 inch diameter were scanned on a high resolution micro-CT at 19µm per
voxel. The purpose was to:
1. Identify rock types and characterise their distribution and volumes within the core
plugs.
2. Select locations for cutting sections through the core plug. Thin sections were
prepared and BSEM images were acquired (see Figure 5). These images were used to
obtain detailed information on mineralogy, grain size, shape and sorting, pore sizes
and any type of clay and cementing minerals.
3. Select location for coring a sub-plug with an attempt to capture a representative
elementary volume (REV) from each core plug. Micro-CT scanning of the sub-plugs
was carried out on a high-resolution scanner, FEI HeliScan (see Figure 5).
 SCA2016-012 5/12

Figure 5: Images of; the core plugs from micro-CT (left), the trim end from BSEM after
segmentation (centre), and the sub-plugs from micro-CT. (Yellow color is quartz; orange is feldspar,
brown is clay and blue is pore)

Modelling
The BSEM images of the thin section of the core plugs were segmented to produce
ternary images of pores; micro-porous phase (clay) and matrix. The ratio of the number
of pixels in each threshold bin to the total number of pixels in the image is an estimate for
the fraction of the mineral in the rock. For each core plug, one PBM was generated,
representing the rock type (RT1) observed in the trim end. For each sample, the
proportion of intergranular and micro-porosity, clay content and PBM size and resolution
and sub-plug image resolution are given in Table 2.
 SCA2016-012 6/12

Table 2: Intergranular and micro-porosity and models size and resolution

Intergranular Micro- PBM size PBM resolution Sub-plug
Sample Clay (%)
porosity (frac) porosity (frac) (µm) (µm) resolution (µm)
Plug 1 0.206 0.039 6.5 1800 1.0 2.09
Plug 2 0.092 0.080 15.8 720 0.4 1.5
Plug 3 0.248 0.025 6.0 2520 1.4 5.12

Pore-scale resolution 3D models were constructed by PBM, based on the information

extracted from the BSEM images. The algorithms applied for the construction of
numerical 3D reservoir rock models are based on simulation of the geological processes
by which the rock was formed: sedimentation, compaction, and diagenesis [13]. There
are several options to place the clay regarding distribution and shape. It can vary from
pore lining clay (as often observed in these samples) to more clustered pore-filling clay
and to clay from feldspar dissolution. The resulting 3D models are quality checked both
visually and statistically against the BSEM images. Figure 6 shows a 3D view of the
cubic PBMs for all three plugs.

Figure 6: 3D image of process based model for each of the three plugs

The 3D micro-CT scans of the core plug were used to visualize the heterogenity and to
select the proper location for one or more subplugs. Micro-CT scan of all three plugs
show that one sub-plug is enough to capture the heterogeneity of the plug. To find an
REV size for the sub-plug, both pore throat and grain size distribution were considered.
BSEM images of the trim ends give useful information about grain size distribution. The
minimum diameter for sub-plug is bigger than 15 times the size of the largest grain
observed in BSEM. Pore throat size distribution from MICP experiment (if it is available)
shows what image resolution is required to be able to capture pore scale phenomenas. For
plugs wih smaller pore throat radii, micro-CT scans require higher resolution thus smaller
sub-plug is needed (like plug 2).
The micro-CT images of the sub-plugs were segmented according to X-ray attenuation of
involved minerals to obtain porosity, micro-porous phase (clay) and matrix. The
segmentation was supported by morphological observation. The size of most clay
minerals is below the resolution of the images and in most cases it introduces an artefact.
Figure 7 shows a 3D view of the micro-CT scans of the sub-plugs and the image
dimensions.
 SCA2016-012 7/12

Figure 7: 3D view of sub-plug micro-CT scans

Flow simulation
To perform multiphase flow simulations, the pore space of the reconstructed models is
transformed into simplified pore networks, which are used directly as inputs to a network
model. Petrophysical properties were calculated for each of the pore scale models,
followed by the extraction of their respective pore networks.
The intergranular porosity is given by resolved pores with sizes greater than or equal to
the image resolution. The total porosity is the summation of intergranular porosity and
sub-resolution porosity (i.e. 0.5-0.6 times the percentage of clay in the rock model). A
Lattice-Boltzmann method is applied to solve Stokes’ equation in the uniform grid model
with maximum 50000 iterations and convergence criteria of 0.005. Flow is driven by a
constant pressure gradient through the model. Permeability is calculated with a fluid that
does not interact with the rock matrix.
For all samples, calculated porosity is similar to helium porosity and calculated absolute
permeability is comparable with Klinkenberg corrected air permeability (see Table 3).
Table 3: Experimental and DRA calculated petrophysical properties
Helium porosity, DRA porosity, Klinkenberg DRA
Sample
(frac) (frac) permeability, mD permeability, mD
Plug 1 0.233 0.245 873 1070
Plug 2 0.173 0.172 1.5 0.82
Plug 3 0.289 0.273 2162 1566

The extracted pore networks were used for the simulation of two-phase fluid flow and the
calculation of relative permeabilities and capillary pressure curves. The following flow
simulations were performed:
• Water-gas primary drainage to initial water saturation (Swi), which is constrained by
a maximum capillary pressure of 12.6 bar.
• Water-gas imbibition to residual gas saturation (Sgrw).
The waterflooding process was simulated by assuming the system is strongly water-wet,
water index=1.0. The advancing contact angle for water phase was assumed 10-35°.
In all simulations, it is assumed that capillary forces dominate (Nca < 1.0E-6). A detailed
account of the methods used for multi-phase flow simulation is given in [14]. Fluid
 SCA2016-012 8/12

injection is simulated from one side of the model (usually the x-direction). Thus, the
entry pressure is a function of the pore sizes present at the inlet.

Results
Plug 1 is moderately homogeneous and poorly consolidated. Simulated capillary pressure
for plug 1 based on the sub-plug and PBM is compared with experimental MICP for a
quality control check in Figure 8-left. Capillary pressure corresponding to 2.09 µm
resolution in sub-plug is 51675 (Pa) at Sw=0.25. For Sw greater than 0.25, capillary
pressure based on the sub-plug is in excellent agreement with the MICP and very good
for the PBM (see Figure 8-left). For Sw less than 0.25, there is an insufficient resolution
to resolve pores invaded by mercury.
Water-gas imbibition relative permeability simulation results for the PBM and sub-plug
are compared with steady state experiments in Figure 9. Simulated water relative
permeability for the sub-plug is similar to the steady state experiment but there is a big
discrepency in the gas relative permeability.

Figure 8: left) Comparison of experimental MICP data with simulated capillary pressure for the
PBM and sub-plug, right) pore size distribution from MICP experiment

Figure 9: Imbibition water-gas relative permeability curves for the PBM, sub-plug, experimental and
Sendra simulation for plug 1
Table 4: Two phase flow results for plug 1
Experiment Experiment
Parameter Sendra PBM Sub-plug
after bump before bump
Swi 0.13 0.13 0.13 0.17 0.21
Sgrw 0.28 0.28 0.19 0.32 0.35
Krw at Sgr 0.105 0.090 0.165 0.187 0.095
 SCA2016-012 9/12

Unlike plugs 1 and 3, plug 2 is tight with well sorted, fine grains and is clay rich. The
porosity of the plug is 0.17 and almost half of it is micro-porosity (0.08). Experimental
MICP pore throat size distribution shows the pore throats are smaller than 1 µm.
segmentation of the micro-CT image of the sub-plug (1.5 µm resolution), resulted in a
high degree of uncertainity for distinguishing between pores and micro-porous clay
patches. As shown in Figure 10, the simulated capillary pressure for the sub-plug is much
lower than the experimental MICP while capillary pressure based on the PBM is in
excellent agreement with the MICP.
The initial water saturation in the sub-plug simulation is very high, Swi=0.48, and so
there is no basis to compare simulated gas-water relative permeability from the sub-plug
with the experiment. However, simulated gas and water relative permeability curves and
residual gas saturation in the PBM are comparable with the steady state experiment
before the bump (see Figure 11).

Figure 10: left) Comparison of experimental MICP data with simulated capillary pressure for the
PBM and sub-plug, right) pore size distribution from MICP experiment

Figure 11: Imbibition water-gas relative permeability curves for the PBM, sub-plug, experimental
and Sendra simulation for plug 2
Table 5: Two phase flow results for plug 2
Experiment Experiment
Parameter Sendra PBM Sub-plug
after bump before bump
Swi 0.24 0.24 0.24 0.18 0.48
Sgrw 0.17 0.26 0.23 0.28 0.33
Krw at Sgr 0.125 0.051 0.056 0.059 0.052
 SCA2016-012 10/12

Plug 3 is an unconsolidated sandstone and the most porous and permeable sample. Due to
the coarseness of the grains, it is difficult to find a REV for a sub-plug. To capture
different features, a bigger sub-plug was required compared to the other two samples (see
Figure 7). For the bigger sub-plug, the resolution of the micro-CT image is lower, 5.12
µm, thus it is not possible to capture the pores smaller than 5.12 µm. As shown in Figure
12, simulated capillary pressure based on the sub-plug is lower than experimental MICP,
especially at lower water saturation. By contrast the PBM simulated capillary pressure is
very similar to MICP. Simulated gas and water relative permeability and residual gas
saturation based on the PBM is comparable with the steady state experiment before
flooding the core plug with high flow rate (see Figure 13).

Figure 12: left) Comparison of experimental MICP data with simulated capillary pressure for the
PBM and sub-plug, right) pore size distribution from MICP experiment for plug 3

Figure 13: Imbibition water-gas relative permeability curves for the PBM, sub-plug and experiment
Table 6: Two-phase flow results for plug 3
Experiment Experiment
Parameter PBM Sub-plug
after bump before bump
Swi 0.11 0.11 0.05 0.12
Sgrw 0.12 0.31 0.32 0.30
Krw at Sgr 0.325 0.131 0.226 0.198

Discussion of Results
For plug 1, the DRA relative permeability on both the sub-plug and PBM show a large
degree of similarity with the experimental relative permeability. For plugs 2 and 3, the
DRA relative permeability based on the PBM is in good agreement with that from the
steady state experiment.
 SCA2016-012 11/12

Comparisons of fractional flow curves of experimental and simulated PBM results

presented in Figure 14 show differences that range from 1 to 4 saturation unit for
recovery at breakthrough (before bump in the experiment). This range is acceptable
considering the fact that the initial water saturations are different in the simulations and
experiments. With respect to the gas relative permeability, however, there is a big
discrepancy in the measured and simulated data. This discrepancy may be due to the fact
that there is a suppression of gas connectivity/conductivity due to non-uniform saturation
profiles (lower gas saturation at the inlet) shown in Figure 2.

Figure 14: Fractional flow curves for experiment and PBM simulation for all plugs

CONCLUSION
Three sandstone samples were tested to compare gas-water relative permeability data
from two different approaches: steady state experiment and DRA. The DRA simulations
were performed on the micro-CT image of the sub-plug and on a process-based model.
Comparison of fractional flow curves from experiment and DRA shows a good
agreement in recovery after breakthrough. The discrepancy in water relative permeability
is within an acceptable range and with one exception, the same trend was observed for
gas relative permeability.
Non-uniform water saturation along the core plug displays artefacts (disequilibria effects
or capillary-end effects) in the coreflood experiments, which can significantly influence
the computation of end-point relative permeabilities and saturation. Numerical
approaches (like Sendra in this work) can be useful to correct for this artefact, but it
doesn’t account for the core heterogeneity and also it doesn’t give a unique result.
In conclusion, there are complications and uncertainties in both experimental and
simulation approaches, but the observed discrepancies have little implications for
practical applications.
Significant improvements have been made in sample handling and thin section
preparation, the micro-CT imaging technology and image processing tools since this
work was completed in 2015.

ACKNOWLEDGEMENTS
We would like to acknowledge BG UK for providing experimental data and allowing the
publishing of the results.
 SCA2016-012 12/12

REFERENCES
[1] A. L. Chen and A. C. Wood, "Rate effects on water-oil relative permeability,"
Society of Core Analysts, 2001, SCA2001-19.
[2] R. Gupta and D. Maloney, "Application of the intercept method to correct steady-
state relative permeability for capillary-end-effect," Society of Core Analysts, 2015,
SCA2015-01.
[3] S. Qadeer, K. Dehghani, D. O. Ogbe and R. D. Ostermann, "Correcting oil/water
relative permeability data for capillary end effect in displacement experiments,"
Society of Petroleum Engineers, 1988, 7423-MS.
[4] F. Hussain, Y. Cinar and P. Bedrikovetsky, "A semi-analytical model for two phase
immiscible flow in porous media honoring capillary pressure," Transport in Porous
Media, 2012, vol. 92, no. 1, pp. 187-212.
[5] M. H. Krause, J. C. Perrin and S. M. Benson, "Modeling permeability distributions
in a sandstone core for history matching coreflood experiments.," Society of
Petroleum Engineers, 2009, SPE 126340.
[6] M. Piri and M. Blunt, "Three-dimensional mixed-wet random pore-scale network
modeling of two- and three-phase flow in porous media. II. Results," PHYSICAL
REVIEW E 71, 2005, vol. 026302, pp. 1-11.
[7] M. Blunta, B. Bijeljica, H. Dongb, O. Gharbia, S. Iglauerc, P. Mostaghimia, A.
Palusznya and C. Pentlanda, "Pore-scale imaging and modelling," Advances in
Water Resources, 2013, vol. 51, p. 197–216.
[8] B. Raeesi, N. Morrow and G. Mason, "Pore network modelling of experimental
pressure hysteresis relationships," Society of Core Analysts, 2013, SCA2013-15.
[9] A. Aghaei and M. Piri, "Direct pore-to-core up-scaling of displacement processes:
Dynamic pore network modeling and experimentation," Journal of Hydrology, 2015,
vol. 522, p. 488–509.
[10] C. H. Pentland, Y. Y. Tanino, S. Iglauer and M. Blunt, "Capillary trapping in water-
wet sandstones: coreflooding experiments and pore-network modeling," Society of
Petroleum Engineers , 2010, SPE 133798.
[11] P. Valvatne and M. Blun, "Predictive pore-scale modeling of two-phase flow in
mixed wet media," Water Resources Research, 2004, vol. 40, pp. W07406-1-21.
[12] F. Lomeland, E. Ebeltoft and T. Hammervold , "A new versatile relative
permeability correlation," Society of Core Analysts, 2005, SCA2005-32.
[13] P.-E. Øren and S. Bakke, "Process based reconstruction of sandstones and prediction
of transport properties," Transport in Porous Media, 2002, pp. 311-343.
[14] P.-E. Øren, S. Bakke and O. J. Arntzen, "Extending predictive capabilities to
network," SPE Journal, 1998, pp. Vol 3, Issue 4.
 SCA2016-020 1/12

COLLABORATION BETWEEN DIGITAL ROCK

ANALYSIS AND LABORATORY FOR GENERATION OF
MULTIPHASE TRANSPORT PROPERTIES FOR
RESERVOIR SAMPLES

Nasiru Idowu1, Alessio Arena1, Ben Young1, Arjen Mascini1, Kurdistan Chawshin1,
Carley Goodwin1, Silvano Sommacal1, Stig Bakke1, Mark Knackstedt1, Pål-Eric Øren1,
and Mohammad Piri2
1
FEI Oil and Gas; and 2 University of Wyoming

This paper was prepared for presentation at the International Symposium of the Society of Core
Analysts held in Snowmass, Colorado, USA, 21-26 August 2016.

ABSTRACT
We present a collaborative study between FEI digital rock analysis (DRA) services and
University of Wyoming on poorly sorted clastic reservoir core materials. The
methodology highlights a multiscale approach which features plug/sub-plug scale
imaging, and submicron resolution using Backscattered Scanning Electron Microscopy
(BSEM) imaging. Pore network models were generated from micron-scale sub-plug
images for the resolved porosity and with a process-based reconstruction method (PBM)
for the sub-resolution porosity. The pore-scale advancing contact angles were measured
in situ and used as inputs in the simulator. The simulation results obtained from the
different pore networks were upscaled using a steady-state technique for each sample.
This approach enables us to generate reliable multiphase transport properties on three bi-
modal sandstone reservoir samples. Simulated results are compared with available
experimental data.

INTRODUCTION
In the current seemingly unending oil price slump, the oil and gas industry must be
innovative and ensure optimal recovery of hydrocarbons from conventional and
unconventional reservoirs at minimal costs. This requires a good understanding of the
reservoir wettability, reliable static/multiphase transport properties and detailed reservoir
characterizations, simulations and management. For gas/water systems [1] and oil/water
water-wet cases [2, 3], DRA offers a timely alternative approach to obtain essential
multiphase transport data from 3D images of rock samples and/or from simplified pore
networks extracted from high resolution images. It also provides the opportunity for fast
and meaningful sensitivity studies on discrete and homogeneous core material.

However, most reservoirs are neither water-wet nor oil-wet and determination of the
actual wettability distribution is crucial [4]. Amott [5] or USBM [6] tests can provide

1
 SCA2016-020 2/12

quantitative averaged wettability information of a core sample. Contact angle, however,

is the most universal measure of the wettability of surfaces [4]. Andrew et al. [7] recently
presented a method for contact angle measurement from X-ray micro-computed
tomography (MCT) of a supercritical CO2-brine system in a Ketton limestone with
simple mineralogy (comprising 99.1% calcite and 0.9% quartz). Qualitative pore-scale
distribution of wettability has also been obtained by integrating information from MCT,
Field Emission Scanning Electron Microscopy (FESEM), and Quantitative Evaluation of
Minerals by SEM (QEMSCAN) [8]. With detailed information on the pore-scale
distribution of wettability [8] or the pore-scale distribution of contact angles [7, 9],
reliable multiphase flow properties can be generated through DRA for non-water-wet
cases.

In this paper, we present a collaborative study between FEI DRA services and University
of Wyoming state-of-the-art laboratory to generate single- and multiphase transport
properties for three bi-modal sandstone reservoir samples. The pore-scale advancing
contact angles were measured in situ and ranged from 90o to 128o using similar procedure
to those stated in [9]. These were used as inputs in a quasi-static pore network simulator
[3] and by so doing, we reduced uncertainties associated with wettability characterization
at the pore-scale and thereby improved the reliability of the predicted data. We explored
one of the strengths of DRA further and generated sensitivity analysis of the predicted
data to different scenarios that would otherwise, have taken years to accomplish in the
laboratory. Finally, we compared simulated results with available experimental data.

IMAGE ACQUSITION AND ANALYSIS

In the original program, twelve conventional plugs were extracted from unpreserved core
material of a clastic reservoir and were characterized texturally and mineralogically. They
were imaged by MCT using a HeliScan micro-CT system. Three of these plugs were
chosen for the study of multiphase flow properties based on their importance to the
reservoir, their specific rock typing and the quality of the reservoir material. An analysis
workflow incorporating conventional and imaging laboratory techniques with DRA
techniques was designed to obtaining the desired results in a timely manner, while
optimizing the costs. A schematic diagram of this workflow is shown in Figure 1.

The three chosen plugs - most relevant to the reservoir-planning scenario - exhibit
different depositional facies, degrees of heterogeneity, different sorting and grain size.
Differences are also noted when comparing measured RCA data: porosities and
permeabilities range from 21% and 768mD for Plug 3 to 28% and 16.8mD for Plug 1.
Petrographically, Plug 1 is a poorly sorted sandstone/siltstone (fine silt to very fine sand)
with abundant microporous pore filling material – Figure 2. The microporous phase is
dominated by micro crystalline quartz and illite. The Backscattered Electron (BSE)
images indicate that a significant part of the porosity is associated with the microporous
phase.

2
 SCA2016-020 3/12

Figure 1. Analysis workflow used for this study

Figure 2. BSE image of plug 1, side length is 1 mm. Right: detail of BSE image and QEMSCAN map
showing the mineralogical nature of the pore filling material and framework grains

Plug 2 is a poorly sorted sandstone/siltstone (medium silt to medium sand) with medium
to poorly rounded grains with some pore filling microporous phase – Figure 3. The
microporous phase consists of kaolinite booklets, micro crystalline quartz, illite and some
carbonate/pyrite.

Plug 3 is a sandstone with a distinct unimodal grain size distribution – Figure 4 and
Figure 7. Medium to coarse well rounded sand grains are surrounded by coarse silt to fine
sand infill. Small amounts of microporous phase are dominated by kaolinite booklets.

3
 SCA2016-020 4/12

Figure 3. Left: BSE image of plug 2, side length is 2.5 mm. Right: Pore filling material

Figure 4. Left: BSE image of plug 3, side length is 5mm. Right: Pore filling kaolinite booklets

The textural differences of those samples led us to use specific coring diameter when
extracting representative sub-plugs. This decision was taken in the attempt to capture a
representative elementary volume (REV) estimated as 15 times the largest observed grain
while maximizing the imaging resolution. Table 1 summarizes the coring diameter,
length and image voxel size for each plug. Acquired images were segmented into X-ray
distinct phases (up to 8 identified phases) using codes based on the converging active
contour algorithm [10]. The dimensions of those images are approximately 1500 voxels
in the x and y direction, and 3000 to 6000 voxels in the z direction; the file size is
approximate 16 to 32 GB of data per image. One vertical slice for each of the three
samples is shown in Figure 5. Table 2 describes the partitioning of the 3D image into
porosity, microporous pore filling material, mineral grains and high density inclusions.
For each of the samples, we extracted an additional 8mm sub-plug used to generate
laboratory MICP data.

4
 SCA2016-020 5/12

Table 1: Sub-plugs extraction/coring diameter, length and image voxel size for each sample
Sample Coring diameter Length (mm) Voxel size (µm)
(mm)
Plug 1 3 9 2.2
Plug 2 4 8 2.9
Plug 3 8 16 5.5

Sub-plugs were subsequently embedded in resin and then cut to prepare a polished
section. This was imaged and analyzed by electron microscopy techniques such as BSE
and Secondary Electron (SE) high magnification imaging. The mosaics of images and
mineralogical maps produced in this way were registered back to the tomogram using a
proprietary algorithm [11]. This enables one to directly quality check (QC) the image
quality and segmentation undertaken on the 3D tomogram to higher resolution BSE data.
In Figure 6, a segmented dataset (3 micron voxel sizes) is registered to a BSE image (500
nm pixel size). The segmentation into 8 phases undertaken in the 3D tomogram correlates
well to solid, microporous and open porosity regions observed in the BSE images.

Table 2. Results from the segmentation of X-ray distinct phases performed on sub-plug µCT images

RESULTS
Single-Phase Properties and MICP
We compute absolute permeability via imaged-based single-phase lattice Boltzmann
permeability simulation. This technique only considers flow along the resolved pore
pathway - through the macropores in the image; the percolation condition was met for all
three sub-plugs based on the careful choice of resolution to target for our X-ray
tomograms. Estimated porosities and calculated absolute permeabilities were then
compared to the experimental laboratory measurement to check the relevance of the sub-
region chosen compared to the original plug scale. Comparison of the experimentally
derived data and image based results are shown in Table 3.

5
 SCA2016-020 6/12

Figure 5. Vertical slices extracted from the sub-plug tomograms, plus zoom in of the highlighted
region in red. The scale represented is valid only for the vertical slice. On the right: grain size
distribution of the three samples calculated from X-ray tomograms as radius of the equivalent sphere

Figure 6. Visual comparison between a BSE image of the polished section prepared within the field of
view of the tomogram and the corresponding horizontal slice from the segmentation result

Table 3. Comparison between laboratory measurement and digital results for porosity and absolute
permeability for all three samples
Porosity (fraction) Absolute permeability (mD)
Sample RCA - plug MICP – sister sub- µCT - sub-plug RCA - Plug µCT - sub-plug
plug
Plug 1 0.282 0.278 0.246 16.8 62
Plug 2 0.293 0.267 0.287 210 225
Plug 3 0.208 0.18 0.175 786 660

6
 SCA2016-020 7/12

Differences in the RCA plug and µCT sub-plug porosity varies from 0.06 to 3.6 porosity
units. Plug 1 has the highest difference in porosity and a factor of approximately 3.7
between the measured (on plug) and computed permeability (on sub-plug). These
differences may be due to highly heterogeneous nature of Plug 1. Measured and
computed values for plugs 2 and 3 agree reasonably well. While a QC for prediction of
permeability requires one to resolve only the largest hydraulic path, other properties like
resistivity, formation factor and two phase properties require one to resolve most of the
accessible pore systems. In Figure 7 we report MICP measurements performed on sister
plugs to the samples used in the imaging process. The grey vertical lines represent the
approximate resolution of the 3D tomograms. This illustrates the porosity that is
identified/resolved in the tomogram and is transformed into simplified pore networks
[12], which are used directly as inputs to a quasi-static flow simulator [3] for each plug.

Figure 7 also illustrates the significant contribution of the microporous regions to the total
porosity. The results mirror the image data acquired in the previous section-- one cannot
directly resolve a significant proportion of the porosity in the plugs 1 and 2 from
tomography alone and one must incorporate information about the microporous regions in
any forward modelling program. For these sub-resolution regions, tomography and
segmentation information enables one to identify the spatial distribution of the microporous
phase in 3D. The nature of this microporous material is then determined using the high
resolution BSE information in (Figure 2 - Figure 4) to develop process and statistically-
based methods to mimic the 3D structure of the microporous zones in the image. The PBM
algorithms are based on simulation of the geological processes by which the rock was
formed; sedimentation, compaction, and diagenesis [13]. Pore networks are also generated
for these sub-resolution pores.

Primary Drainage and Waterflooding Simulation Results

Before simulating oil/brine displacement processes and generating the desired transport
properties, we first establish for each rock sample whether the extracted pore networks
from MCT images and PBM are representative or not. This is achieved by simulating
oil/brine primary drainage on the pore networks and then upscaling [14] the resulting
network-predicted oil/brine Pc – Sw curves from the different networks for each sample.
The upscaled oil/brine Pc – Sw curve for each sample is then compared with measured
mercury capillary pressure data on sister samples as shown in Figure 8 for Plug 2. The
experimental capillary pressure data are scaled to oil/brine Pc – Sw curves using interfacial
tensions and contact angles in Table 4. The match between the simulated and measured
data is good and confirms that the different pore classes are well captured and the networks
are representative.

The oil/brine primary drainage capillary pressure simulation results shown in Figure 8 were
stopped at 10 bar to match the entire MICP curves. However, based on the estimated
capillary pressure values using oil and brine density and height above the oil-water-contact,
the primary drainage simulations were re-run and stopped at a maximum capillary pressure
of 2 bar and then followed by waterflooding. The average pore-scale advancing contact

7
 SCA2016-020 8/12

angle of 109o measured in situ at University of Wyoming, USA was used as input [5]. For
this case, we assume 80% of all pores/throats invaded by oil during primary drainage
changed wettability from initial water-wet state to oil-wet condition and investigate
sensitivity analysis of the results to different percentages. Oil-wet advancing contact angles
for the oil-wet pores/throats are distributed between 100o to 118o. For the remaining water-
wet pores/throats, the advancing angles are distributed between 30o to 70o. Figure 8 depicts
the primary drainage (PD) and waterflooding (WF) capillary pressure curves and relative
permeability results for Plug 2. The end-point saturations and relative permeability values
for all the three samples are summarized in Table 5.

Figure 7 MICP measurement for sister plugs of the three samples studied. The black curve
represents the pore size distribution, the red is the cumulative one. The vertical line is the
approximate resolution of our tomograms.

Table 4: Input data to flow simulations

Water density [kg/m3] 1000
Oil density [kg/m3] 700
IFToil/brine [dynes/cm] 30
IFTHg/air [dynes/cm] 480
Water-wet receding angles [degr.] 0° – 10°
Water-wet advancing angles [degr.] 30° – 70°
Oil-wet advancing angles [degr.] 100° – 118°
Mercury/air contact angles [degr.] 140°
Percentage of oil-wet
80
pores/throats

8
 SCA2016-020 9/12

Table 5: Measured and simulated end-point saturations and relative permeability values
Simulated
Property Measured
Sample 1 Sample 2 Sample 3
Porosity (fraction) 0.27 0.24 0.25 0.19
Permeability (mD) 130 – 160 93 286 814
Swi (fraction) 0.16 0.18 0.15 0.04
Kro@Swi 0.74 0.98 0.99 0.99
Sorw 0.26 0.32 0.28 0.20
Krw@Sorw 0.28 0.25 0.45 0.40

Figure 8: capillary pressure and relative permeability curves for primary drainage (top) and
waterflooding (bottom) for Sample 2.

9
 SCA2016-020 10/12

Comparison of Simulated Results to Measured Data and Sensitivity Analysis

Capillary pressure and relative permeability data were not measured during the
experiments at University of Wyoming. However, porosity, permeability, end-point
saturations and end-point relative permeability to water at residual oil saturation were
measured on a preserved sample adjacent to sample 2 plug using unsteady-state method.
Figure 8 shows the measured end-point values while Table 4 compares the simulated
results with measured data. There are good agreements between measured and simulated
initial water saturation and residual oil saturation despite the differences in porosity and
permeability values. However, lower measured end-point relative permeability values
may be due to uncertainties in measured absolute permeability values (130 -160 mD).

DRA enables one to investigate the sensitivity of the simulated results to the percentage
of all pores/throats that changed wettability from initial water-wet state to oil-wet
condition. We investigated this by running simulations for completely water-wet and oil-
wet cases. The percentage of all pores/throats that changed wettability from initial water-
wet state to oil-wet condition is 0% and 100% for water-wet and oil-wet case
respectively. Oil-wet advancing contact angles for the oil-wet pores/throats are
distributed between 140o to 170o for the completely oil-wet case while the advancing
angles for the completely water-wet case remained the same (30o to 70o). The results of
the sensitivity analysis compared with the waterflooding results presented in the previous
subsection (base case) are shown in Figure 9.

Discussion of Results
The field scale implications of the relative permeability curves depicted in Figure 9 in
terms of recovery is investigated using one-dimensional Buckley-Leverett analysis [15].
Assuming oil/brine viscosity ratio of 2, Figure 10 shows recovery as a function of pore
volume of water injected. This figure highlights the importance of using appropriate
multiphase transport data generated with correct pore-scale distribution of wettability.
This will ensure good planning and forecasting and optimal recovery of hydrocarbons at
minimal costs.

CONCLUSIONS
Uncertainties associated with wettability characterization at the pore-scale during DRA
can be eliminated by the use of in situ contact angle measurements. Integration of
detailed information on the pore-scale distribution of contact angles with topologically
equivalent networks can guide network modelling of multiphase transport properties and
considerably improve reliability of predicted results. For non-water-wet systems, we
advocate an analysis workflow that incorporates conventional and imaging laboratory
techniques with DRA techniques for generation of desired multiphase transport
properties.

10
 SCA2016-020 11/12

Figure 9: Capillary pressure and relative permeability sensitivity analysis waterflooding results for
oil-wet (WF_OW), base case – mixed-wet (WF_MW) and water-wet (WF_WW)

Figure 10: Oil recovery in pore volume (PV) versus PV of water injected using 1-D Buckley-Leverett
analysis for the relative permeability curves depicted in Figure 9

ACKNOWLEDGMENTS
The authors acknowledge FEI Oil and Gas Business for permission to publish this paper
and Hess Corporation, USA for allowing us to publish these data on their samples.

REFERENCES
1. Farokhpoor, R., Westphal, E., Idowu, N., et al., “Gas-water steady-state relative
permeability determination with two approaches: experiment and digital rock
analysis, strengths and weaknesses”, Paper SCA2016-Temp_137, Proceedings of
the 2016 SCA Symposium, Snowmass, Colorado, USA, 21 - 26 August, 2016.

11
SCA2016-020 12/12

2. Bakke, S., and P.E. Øren, “3-D pore-scale modelling of sandstones and flow
simulations in the pore networks,” SPE Journal, (1997), 2, 136-149.
3. Øren, P.E., S. Bakke, and O.J. Arntzen, “Extending Predictive Capabilities to
Network Models”, SPE Journal, (1998), 3, 324-336.
4. Morrow, N.R., “Wettability and its Effects on Oil Recovery”, J. Pet. Tech. (1990),
42: 1476-1484.
5. Amott, E., “Observations Relating to the Wettability of Porous Rock”, Trans.
AIME, (1959), 216: 156-162.
6. Donaldson, E.C., Thomas, R.D. and Lorenz, P.B., “Wettability Determination and
its Effect on Recovery Efficiency”, SPE Journal, (1969) 9: 13-20.
7. Andrew, M., Bijeljic, B., and Blunt, M.J., “Pore-scale Contact Angle
Measurements at Reservoir Conditions Using X-ray Microtomography”,
Advances in Water Resources. (2014), 68: 24-31.
8. Idowu, N., Long, H., Øren, P.E., et al., “Wettability analysis using micro-CT,
FESEM and QEMSCAN and its applications to digital rock physics”, Paper
SCA2015-010, Proceedings of the 2015 SCA Symposium, St. John’s
Newfoundland and Labrador, Canada 16-21 August, 2015.
9. Aghaei, A., and Piri, M., “Direct pore-to-core up-scaling of displacement
processes: Dynamic pore network modelling and experimentation”, Journal of
Hydrology, (2015), 522, 488-509.
10. Sheppard, A.P., Sok, R.M. and Averdunk, H., “Techniques for Image
Enhancement and Segmentation of Tomographic Images of Porous Materials”,
Physica A (2004), 339 (1-2), 145-151.
11. Latham, S, Varslot, T and Sheppard, A., “Image registration: Enhancing and
Calibrating X-ray Micro-CT Imaging”, Paper SCA2008-35, Proceedings of the
2008 SCA Symposium, Abu Dhabi, UAE, 29 October – 2 November, 2008.
12. Idowu, N., Nardi, C., Long, H., et al., “Pore-Scale modelling: Effects of network
properties on predictive capabilities”, Paper SCA2012-35, Proceedings of the
2012 SCA Symposium, Aberdeen, Scotland, UK, 27-30 August, 2012.
13. Øren, P.E. and Bakke, S., "Process Based Reconstruction of Sandstones and
Prediction of Transport Properties", Transport in Porous Media, 2002, 46, 311-
343
14. Lopez, O., Mock, A., Øren, P.E., et al., “Validation of Fundamental Carbonate
Reservoir Core Properties Using Digital Rock Physics” Paper SCA2012-19,
Proceedings of the 2012 SCA Symposium, Aberdeen, Scotland, UK, 27-30
August, 2012.
15. Buckley, S.E., and Leverett, M.C., “Mechanisms of fluid displacement in sands”,
Trans. AIME, (1942), 146, 107–116

12
 SCA2016-003 1/12

AN IMPROVED INSIGHT INTO LOW-SALINITY

WATERFLOODING: IN-SITU CHARACTERIZATION OF
WETTABILITY ALTERATION AT ELEVATED
PRESSURE AND TEMPERATURE CONDITIONS
M. Khishvand, A.H. Alizadeh, I. Oraki Kohshour, M. Piri
Department of Petroleum Engineering, University of Wyoming, Laramie, WY 82071, USA
This paper was prepared for presentation at the International Symposium of the Society of Core
Analysts held in Snowmass, Colorado, USA, 21-26 August 2016

ABSTRACT
Low-salinity waterflooding (LSWF) is known as one of the most effective improved oil
recovery techniques that could result in significant additional recovery compared to
conventional high-salinity waterflooding (HSWF). Although numerous laboratory studies
have confirmed the effectiveness of LSWF under specific conditions, they have mostly
failed to present explicit evidences on the pore-level mechanisms that are responsible for
the observed improvement in recovery. In this study, we investigated LSWF production
mechanisms using X-ray micro-computed tomography (micro-CT) and examined
recovery trends at the pore scale. Two core-flooding experiments were performed on
miniature reservoir sandstone core samples at elevated pressure and temperature
conditions. The preserved core samples (5 mm in diameter) were cut and then saturated
to establish reservoir initial saturation conditions. The samples were subsequently
waterflooded with low- and high-salinity brines and imaged using a micro-CT scanner.
Micro-CT images were then used to obtain fluid saturations and three-dimensional maps
of fluid occupancy during each experiment. High-resolution micro-CT images were also
analyzed to measure pore-scale contact angles and study in-situ wettability and its impact
on oil mobilization during different stages of waterflooding process.

The results highlight a significantly improved performance in LSWF compared to that of

HSWF. The LSWF test showed more prolonged oil recovery response and gradual
recovery at later stages while the HSWF recovery stabilized much earlier. Pore-scale
contact angle measurements yielded direct evidence of wettability alteration from weakly
oil-wet toward weakly water-wet conditions during LSWF; however, no considerable
changes were observed in in-situ contact angles during the HSWF. Investigation of fluid
occupancy maps along with in-situ characterization of wettability allowed us to establish
a significantly better understanding of pore-scale mechanisms responsible for improved
recover by LSWF.

INTRODUCTION
Waterflooding has been frequently implemented as an improved oil recovery (IOR)
method to maintain reservoir pressure and sweep oil toward production wells. This
technique was first practiced in 1880 [1] and since then has been adopted as one of the
 SCA2016-003 2/12

most widespread methods for oil recovery. Due to the success of this IOR scheme in
numerous oilfields, a large number of studies have targeted different aspects of
waterflooding. Yildiz and Morrow [2] found that reservoir wettability and composition of
injected and connate brines considerably affect waterflood recovery. These findings
immediately inspired scientists to focus on composition and salinity of injected brine as
the only controllable parameters to obtain higher waterflood efficiency. Primary research
on outcrop Berea sandstone samples revealed that reducing salinity of the injected brine
had a substantial effect on the amount of trapped oil saturation [3 and 4]. This method, so
called low-salinity water flooding (LSWF), soon became one of the most promising IOR
techniques. It was observed that LSWF resulted in remarkably higher recoveries in both
sandstone [5] and carbonate [6] rock samples. In some cases, single well tracer tests [8]
proved the favorable laboratory results. LSWF can be performed under secondary mode
(at initial water saturation) or tertiary mode (at waterflood residual oil saturation). Either
way, it may offer notably higher recovery efficiency compared to high-salinity
waterflooding (HSWF). During the last decade, numerous laboratory tests have been
conducted to evaluate the effectiveness of LSWF, investigate conditions under which
LSWF succeeds, and describe displacement mechanisms governing LSWF recovery. The
observations have demonstrated that the success of LSWF is greatly influenced by the
original reservoir wettability and fluids present in the pore space (i.e., connate brine and
crude oil). Tang and Morrow [3] reported three conditions that are necessary, but not
sufficient, to observe low-salinity effect (LSE). These include significant clay content,
presence of connate water, and exposure to crude oil to make a mixed-wet condition.
Although some studies have shown the LSE for clay-free sandstone [9] and carbonate
samples [7], these three conditions are universally pervasive to select candidate reservoirs
for LSWF.

To perform successful LSWF, the underlying displacement mechanisms responsible for

LSE must be comprehensively understood. Over the years, many research studies have
been devoted to probe physical phenomena and displacement mechanisms that explain
the higher recovery of LSWF and enrich the fundamental understanding of the LSE; but
still, no consistent and systematic answers have been presented [3]. Results of core-
flooding experiments, such as saturation trends, effluents analysis, and pressure drop data
have been used to provide some indirect evidences supporting mechanisms of the LSWF
process. Most of these studies have discussed several main hypotheses supporting the
observed increases in recovery during LSWF. The most well-known mechanisms are
detachment and migration of mixed-wet clay fines [2] and associate wettability alteration
[3, 10, and 11], multicomponent ion exchange (MIE) between the low-salinity brine
(LSB) and clay surfaces [4 and 12], mineral dissolution [13], and saponification [7].
Among these, wettability alteration (due to limited release of mixed-wet clay fines and/or
MIE) from mixed-wet toward increased water-wetness is one of the most supported and
frequently suggested mechanisms. When an oil-wet rock surface is exposed to LSB,
water destabilizes the clays, acts as a wettability modifier, and changes the wettability
toward increased water-wetness. This indeed allows oil to be displaced from the oil-wet
pores [11]. Although there have been some attempts based on pore-scale theoretical
 SCA2016-003 3/12

models to explain the pore-scale subtleties of this phenomenon [11], no rigorous pore-
scale experimental observations have been presented in this regard, and hence pore-scale
explanation of displacement mechanisms that govern higher recovery of the LSWF
remain as a question that has yet to be answered.

For years, scientists have performed macro-scale tests to probe the above-mentioned
mechanisms, but have not been able to directly observe the rock-fluid interactions, such
as wettability alteration, and dynamic reactive interplays occurring at the pore scale. This,
perhaps is the most significant shortcoming of those studies. Recently, however, image-
based techniques such as micro-computed tomography (micro-CT) imaging has been
employed to eliminate the drawbacks associated with macro-scale tests. Micro-CT
imaging is a technology by which a rock sample is imaged to map its internal
morphology and pore-scale fluid occupancy during different flow processes [14-16].
Using this new technology, one can look inside the pore space during LSWF, monitor
potential changes in wettability and saturation trends, and obtain direct observations of
governing mechanisms for the LSE. A few research groups have attempted to take
advantage of microtomography along with scanning electron microscope (SEM) imaging
technique to observe mineral dissolution and fines migration during LSWF [9 and 13]
and some successes have been achieved. However, due to technical difficulties and
limitations of experimental procedures used, microtomography has not been used to
investigate potential in-situ wettability alteration and its associated displacement events
as, perhaps, the most important mechanism responsible for LSE.

In this work, we endeavoured to exploit micro-scale images generated during LSWF

experiments to investigate wettability alteration and provide observational clues tying the
in-situ wettability with governing displacement mechanisms. An extensive laboratory
study was deployed on reservoir sandstone samples and using reservoir crude oil at
reservoir conditions. We assembled a unique miniature core-flooding apparatus, which
could be employed to perform experiments at reservoir conditions while the core holder
is placed inside a micro-CT scanner and the sample is imaged in the course of the
experiments. We carried out two sets of experiments (one HSWF and one LSWF) to first
examine the efficiency of the LSWF compared to that of HSWF in reservoir sandstone
samples, and secondly to investigate the potential wettability alteration during LSWF in
order to develop a better understanding of the pore-scale displacement mechanisms
responsible for LSE. We used an effective approach to directly measure in-situ contact
angles in individual pores. In-situ wettability data along with spatial distribution maps of
fluid phases were used to shed light on the displacement events responsible for LSE.

Experimental
In this section, we describe the rock samples and fluids used to perform the experiments.
We then discuss the experimental setup utilized to conduct the tests followed by the
experimental procedure.
 SCA2016-003 4/12

Rock Samples
We performed the core-flooding experiments on nominally 5-mm diameter and 50-mm
long miniature reservoir sandstone core samples labeled as Samples A and B. The
samples were taken parallel to bedding from a 110-mm diameter preserved reservoir core
sample. The parent core sample was loosely consolidated and extra effort was made to
drill smooth miniature core samples. Air was used as a coolant to minimize disturbances
to the samples. All samples were dry-cut, instantly placed into the core holder, and then
undergone their corresponding experimental procedure. Additionally, a 25.4-mm core
plug (Sample C) and an end trim (Sample D) were cut from the same parent core. Sample
C was used to measure initial water saturation (Swi) using dean-stark method. Dean-stark
measurement showed Swi of 13% for the preserved core after drilling. This sample was
then solvent-cleaned and its porosity and permeability were measured. Sample D was
used for XRD analysis that showed about 6% clay content, mainly kaolinite. The
presence of connate water and the significant clay content suggested that this rock might
be a good candidate for LSWF. For Samples A and B, porosity was measured using
micro-CT images. Uniform porosity distributions (see Figure 1) verified the homogeneity
of the samples. Geometrical and petrophysical properties of the rock samples are listed in
Table 1.

Fluids
High-salinity brine (HSB) was synthetized using sea water formulation with a total
dissolved solid (TDS) of 40,300 ppm. The synthetic brine was prepared by including
most of the ions in order to replicate the sea water composition as closely as possible.
This brine was then diluted with a ratio of one-fiftieth and utilized as the LSB. The oil
phase was reservoir stock tank oil (i.e., dead crude oil). Crude oil was filtered through a
0.5-µm filter before use to remove residue and debris. We then added 6 vol. %
diiodomethane as X-ray dopant to ensure sufficient contrast between the crude oil and
aqueous solution in X-ray images. Since we used ethane in the course of the experiments,
one could argue possible asphaltene precipitation when the crude oil came into contact
with ethane. To investigate this, we mixed 10 cc of the crude oil with 50 cc of ethane in a
high-pressure cell and frequently shacked the container at experimental conditions (i.e.,
1000 psi and 60 °C) for a day. The fluid mixture was then removed and the precipitated
asphaltene was measured. The test showed 0.29 wt.% of asphaltene precipitation, which
was negligible and did not introduce any adverse uncertainties into the experiments.
Density and viscosity of all fluids were either measured at experimental conditions or
estimated from data available in the literature [18-20]. Fluid properties are summarized in
Table 2.

Experimental Setup
We employed a unique experimental apparatus to carry out the micro-scale core-flooding
tests at elevated pressure and temperature conditions. The core-flooding was integrated
with a high-resolution micro-CT scanner to image the core samples in the course of the
experiments. The experimental setup composed of four dual cylinder QuizixTM pumps, a
miniature core holder, a micro-CT scanner, Rosemount pressure transducers, and
 SCA2016-003 5/12

appropriate heating systems. All wetted parts of the core-flooding system were made out
of HastelloyTM, titanium, and other corrosion resistant material. Figure 2 exhibits a
schematic diagram of the flow system used in this study. Pumps 1 and 2 withdrew brine
and oil, respectively, from buckets, and injected them into the core samples. Fluids could
be pressurized with their respective pumps and passed through controllable heat
exchangers to reach desired temperatures prior to injection. Pump 3 received the effluents
at desired constant pressures and kept the pore pressure stable. A net overburden pressure
was maintained during all experiments using Pump 4. The custom-built core holder was
fabricated from composite materials with minimum X-ray absorption. It had nozzles that
directed the fluids from Pumps 1 and 2 into the core and from the core to the back
pressure pump (i.e., Pump 3). A special heating system was coupled with the core holder
and the flooding system to heat up the flooding vessels during the experiments. All lines
were insulated to maintain the fluid temperature and minimize potential flow distortions
due to the changes in the temperature. Further details of the core-flooding system can be
found elsewhere [17].

Experimental Procedure
To conduct each experiment, the core sample was first cut and instantly loaded in the
miniature core holder. A low (for easier access to the pore space) net confining pressure
was applied. The sample was then flooded with ethane to remove bulk air from the pore
space. The ethane flooding was performed at 100 psi back pressure and at moderately
low gas flow rates to minimize the risk of vaporization of the connate brine. The sample
was then scanned to capture a reference image. Afterward, pore pressure was reduced and
the dead crude oil was introduced into the sample. When crude oil breakthrough and
removed some of the bulk ethane, pore and net confining pressures were simultaneously
increased to 800 and 900 psi, respectively. At this step, temperature of the system was
gradually increased to 60 °C. These temperature and pressure conditions were fixed until
the end of the experiment. The oil was allowed to remove ethane thoroughly with a
miscible displacement process. Once the flow of oil was established, oil injection was
continued for a day with a very low flow rate to ensure no ethane was left in the pore
space. The core was frequently scanned for any evidence of hydrocarbon gas. When zero
gas saturation was confirmed by the CT images, a scan was run to determine the initial
water saturation profile along the sample. Thereafter, the core was flooded with desired
brine at a pre-specified flow rate of 0.004 cc/min. The flow rate was equivalent to
capillary numbers of 1.42E-07 and 1.48E-07 for LSWF and HSWF tests, respectively.
Brine was injected with increments of 0.5 pore volume (at the outset of the experiment)
and 1 pore volume (at the end). After each increment, we halted brine injection and the
system was let come to rest. The middle section of the core sample, hereafter referred to
as field of view (FOV), was then scanned. The final stage of each waterflood was
achieved when the fluid occupancy no longer changed even by doubling the brine flow
rate. Image sets were then reconstructed as form of stacked slices and prepared for
analysis.
 SCA2016-003 6/12

IMAGE ANALYSIS
The tomographic images obtained during the experiments were analyzed to map fluid
occupancy and measure trapped oil saturation as well as fluid-rock properties. The image
set captured at each stage of the experiments was segmented after registration with the
reference image. The segmented data was used to characterize in-situ wettability by
means of oil-brine contact angle measurement. This provided a platform to study the
potential wettability alteration in the rock and the displacement mechanisms responsible
for the LSWF recovery. A detailed explanation of the segmentation method and the
procedure for contact angle measurements can be found elsewhere [14-16].

RESULTS AND DISCUSSION

In this section, we present the results of the micro-scale LSWF and HSWF experiments
performed on Samples A and B. For each experiment, we used pore-scale fluid
occupancy maps generated utilizing the micro-CT images to study displacement
mechanisms and resulting saturation trends. This allowed us to highlight the efficiency of
the LSWF compared to that of HSWF in the targeted rock samples. Furthermore, we
characterized in-situ wettability at different stages of each flow process to probe potential
wettability alteration during the LSWF. Finally, we attempted to connect the in-situ
wettability alteration and its consequent pore-scale displacements with the saturation
trends observed.

Pore-Scale Fluid Occupancy and Saturation

Figures 3 and 4 exhibit pore-scale fluid occupancy maps of the core samples subjected to
the LSWF and HSWF, respectively. Each figure contains several images corresponding
to different stages of each flow test and exemplifies the displacement events taking place
in the pore space. As seen in Figures 3a and 4a, the pore space was dominantly occupied
by oil before the commencement of the waterfloods. Oil almost entirely filled the large
and medium-sized (mostly oil-wet) pores. Water, however, resided in very small (mostly
water-wet) pores and crevices. This fluids arrangement established the initial mixed-wet
conditions of the samples. As waterflooding initiated (Figures 3b for the LSWF and 4b
for the HSWF), in both samples, water invaded into fully and partially oil-filled elements
by two mechanisms. It displaced oil through imbibition from water-wet pores (with an
order of smallest to largest) and upon drainage from oil-wet pores (with an order of
largest to smallest). This could be realized when one compares Figures 3b and 4b with 3a
and 4a. Water, initially, invaded into small water-wet pores (squares) and large oil-wet
pores (circles) while medium-sized pores remained oil-filled. In other words, the oil
clusters sitting in the medium-sized pores were the last ones to be mobilized by water. As
waterflood proceeded (Figures 3c and 4c), water had more accessibility to the entrance of
the oil-filled pores and hence, it invaded into a greater number of them including some
medium-sized ones. Meanwhile, oil gets trapped in some pore elements depending on
wettability and fluids present in the neighbouring elements. Finally, the waterfloods
ended at a point that the remaining oil was entirely trapped (i.e., disconnected). This
point, known as waterflood end point, signified the efficiency of the waterfloods.
Comparison between Figures 3c and 3d and 4c and 4d revealed that the end point was
 SCA2016-003 7/12

established after about 5 pore volumes of water injection (PVWI) for the HSWF;
however, in the case of the LSWF, end point was obtained much later and pronounced
changes in oil occupancy were observed even in late stages (until about 10 PVWI).

To determine the end points and examine the efficiency of the LSWF, saturation values
were calculated over the FOVs for both waterfloods. The results are presented in Figure 5
for stages with the most considerable variations. Before waterflooding, initial oil
saturations (Soi = 1 - Swi) were homogenous along the FOV and had very similar average
values for both samples (about 0.91 and 0.90 for Samples A and B, respectively). These
values were also consistent with the Soi measured using dean-stark method for Sample C
(i.e., 87%). After 5 PVWI, oil saturations were reduced in both samples with similar
trends resulting in remaining oil saturations (Sor) of 0.46 and 0.45 for Samples A (LSWF)
and B (HSWF), respectively. This in fact indicated similar initial recoveries of the LSWF
and HSWF. Upon continuation of the waterfloods, for the HSWF, Sor stabilized and no
further considerable oil production was observed, even with an increase in flow rate to
0.006 cc/min; therefore, this point was considered as wateflood residual oil saturation
(Sorw) for the HSWF. In contrast, for the LSWF, Sor continued to reduce significantly at
the later stages of waterflooding (until 10 PVWI) resulting in a Sorw of 0.30. These
observations demonstrated that the LSWF had a considerable gradual recovery at later
stages whereas the high-salinity production stabilized much earlier. These differences in
recovery trends, known as LSE, are in line with our earlier observations of water-
displacing-oil events in medium-sized pores occurred during the late stages of the LSWF
(Figure 3d). We believe that when low-salinity brine contacted the rock surface, it
changed the wettability toward increased water-wetness (see more details listed later in
this section). The wettability alteration might be attributed to either migration of clay
minerals or multiple ion exchange [3 and 5]. LSB gradually altered the wettability of the
rock and as a consequence facilitated invasion of water into medium-sized oil-filled pores
(see the rectangular area shown in Figure 3d). To verify this hypothesis, one would need
to investigate whether wettability alteration does take place during the LSWF and how
the potential alteration impacts oil displacement. In the next section, we explore this by
means of in-situ wettability measurement and pore-scale investigation of displacement
events motivated by potential wettability alteration.

In-Situ Characterization of Wettability Alteration

We accomplished a detailed analysis of the micro-scale images to find potential changes
in wettability during different flooding scenarios. For each flooding step, the segmented
data was analyzed to measure oil-water contact angles. The detailed methodology of the
contact angle measurement can be found elsewhere [15 and 16]. We limited our
measurements to contact angles at main terminal menisci (i.e., oil-water interfaces at the
center of pore-throat junctions). Since we imaged the samples after the fluids were let to
come to rest, contact angle measurements were considered to give equilibrium values.
One should note that it was impractical to measure contact angles in very small (mainly
water-wet) pores because the amount of oil present in those pores was extremely small.
Contact angle measurement results presented in Figure 6 are for three stages of each
 SCA2016-003 8/12

waterflood (i.e., after 0.5, 5, and 10 PVWI). Contact angle values show a distribution for
each stage, which might be attributed to local mineral heterogeneity and microscopic
surface roughness. As it can be observed, initial contact angle values (i.e., after 0.5 PV)
showed similar distributions as well as mean values for the LSWF and HSWF. The
average values of about 115 and 117 degrees for the LSWF and HSWF, respectively,
verified a weakly oil-wet condition in the preserved samples. This figure also
demonstrates that contact angles did not change during the course of HSWF, while
substantial changes toward increased water-wetness occurred during the LSWF (from
115 to 89°). As the LSWF proceeded, LSB may have reacted with the clay minerals
lining the rock surface. It has been hypothesized that LSB may detach the mixed-wet clay
particles from the pore walls [3] and exchange divalent ions (such as Ca2+) with the clays
[5]. Both mechanisms may have contributed to the observed reduction in the oil-water
contact angle. The bimodal graphs shown in Figure 6 (after 5 and 10 PV LSWF) indicate
that contact angle could change significantly in some pores whereas others might show
only slight alterations compared to their initial wettability condition. This might arise
from several key factors such as different contact time and contact areas with the LSB as
well as variable clay content of the pores.

Additional analysis of the pore-scale contact angles could further corroborate the
displacement events that were discussed earlier as a consequence of wettability alteration
during LSWF. Figure 7 depicts two-dimensional views of fluid occupancy in an
individual pore and the corresponding measured contact angles after 0.5, 5, and 10
PVWI. This figure illustrates that water had access to the entrance of an oil-filled element
in all stages, but interestingly, the pore was only invaded by water when the contact angle
decreased. This can be explained by threshold capillary pressure concept [11]. Upon
changes in wettability from 127 (after 0.5 PVWI) to 99 (after 5 PVWI), the threshold
water pressure needed for the drainage process (i.e., water-displacing-oil in this oil-wet
pore) was reduced, which in turn enabled water to proceed in the pore and displace the oil
in the later stages (i.e., between 5 and 10PVWI). We believe that these changes in
wettability and subsequent invasion of water into medium-sized oil-wet pores directed
the gradual production of oil at late stages of the LSWF and were responsible for higher
oil recovery of the LSWF. This study provided, for the first time, direct evidences of
wettability alteration due to LSE and associated pore-scale mechanisms governing the
greater recovery established by LSWF.

CONCLUDING REMARKS
In this study, we performed a LSWF and a HSWF micro-scale experiments on preserved
reservoir sandstone core samples with reservoir fluids and at elevated pressure and
temperature conditions. The preserved samples were cut and saturated using a special
procedure to establish initial reservoir saturation. The samples were then flooded under
secondary mode using brine with different salinities. We imaged the fluid occupancy in
the pore space and used the images to characterize oil saturation and in-situ wettability as
well as probing the physical mechanisms responsible for LSE.
 SCA2016-003 9/12

The results indicated waterflood residual oil saturations (Sorw) of 0.30 and 0.45 for the
LSWF and the HSWF, respectively. These observations highlighted the significant
performance of LSWF compared to HSWF in our study. LSWF showed more prolonged
oil recovery response (gradual recovery at later stages) while the HSWF recovery curve
stabilized much earlier. Analysis of micro-CT images enabled us to measure pore-scale
contact angles to examine wettability alteration of the rock surface. These measurements
coupled with pore-scale fluid occupancy maps revealed that when the rock surface was
exposed to low-salinity brine, the average oil-water contact angle gradually reduced
toward neutral-wet condition (i.e., from 115 to 89), which in turn lowered the threshold
water pressure needed to displace oil in the oil-wet pores. As a consequence, water
invaded into more oil-filled pores and increased recovery at the later stages of the LSWF.

ACKNOWLEDGEMENTS
We gratefully acknowledge financial support of Hess Corporation and the School of
Energy Resources at the University of Wyoming.

REFERENCES
1. Carll J. F., “The geology of the oil regions of Warren, Venango, Clarion and
Butler Counties”, Report III, Harrisburg, Pennsylvania, 1880.
2. Yildiz, H. O., Valat, M., and Morrow, N. R., “Effect of brine composition on
wettability and oil recovery of a Prudhoe Bay crude oil”, Journal of Canadian
Petroleum Technology, (1999), 38, 01.
3. Tang, G. Q., and Morrow, N. R., “Salinity, temperature, oil composition, and oil
recovery by waterflooding”, SPE Reservoir Engineering, (1997), 12, 04, 269-276.
4. Morrow, N. R., and Buckley, J., “Improved oil recovery by low-salinity
waterflooding”, Journal of Petroleum Technology, (2011), 63, 05, 106-112.
5. Lager, A., et al., “Low salinity oil recovery - An experimental investigation”,
Petrophysics, (2008), 49, 01.
6. Fathi, S. J., Austad, T., and Strand, S., “Water-based enhanced oil recovery
(EOR) by “smart water”: Optimal ionic composition for EOR in
carbonates”, Energy & Fuels, (2011), 25, 11, 5173-5179.
7. Jerauld, G. R., et al., “Modeling low-salinity waterflooding”, SPE Reservoir
Evaluation & Engineering, (2008), 11, 06, 1000-1012.
8. McGuire, P. L., et al., “Low salinity oil recovery: An exciting new EOR
opportunity for Alaska's North Slope”, SPE Paper 93903 presented at the SPE
Western Regional Meeting, 30 March-1 April 2005, Irvine, California.
9. Fogden, A., et al., “Mobilization of fine particles during flooding of sandstones
and possible relations to enhanced oil recovery”, Energy & Fuels, (2011), 25, 4,
1605-1616.
10. Austad, T., RezaeiDoust, A., and Puntervold, T., “Chemical mechanism of low
salinity water flooding in sandstone reservoirs”, SPE Paper 129767 presented at
the SPE Improved Oil Recovery Symposium, 24-28 April 2010, Tulsa, Oklahoma.
 SCA2016-003 10/12

11. Sorbie, K. S., and Collins, I., “A proposed pore-scale mechanism for how low
salinity waterflooding works”, SPE Paper 129833 presented at the SPE Improved
Oil Recovery Symposium, 24-28 April 2010, Tulsa, Oklahoma.
12. Lebedeva, E. V., and Fogden. A., “Micro-CT and wettability analysis of oil
recovery from sand packs and the effect of waterflood salinity and
kaolinite”, Energy & Fuels, (2011), 25, 12, 5683-5694.
13. Pu, H., et al., “Low-salinity waterflooding and mineral dissolution”, SPE Paper
134042 presented at the Annual Technical Conference and Exhibition, 19-22
September 2010, Florence, Italy.
14. Alizadeh, A.H., et al., “Multi-scale experimental study of carbonated water
injection: An effective process for mobilization and recovery of trapped oil”,
Fuel, (2014), 132, 219-235.
15. Khishvand, M., Akbarabadi, M., and Piri, M., “Micro-scale experimental
investigation of the effect of flow rate on trapping in sandstone and carbonate
rock samples”, Advances in Water Resources, (2016), in press.
16. Khishvand, M., Alizadeh, A.H., and Piri, M., “In-situ characterization of
wettability and pore-scale displacements during two- and three-phase flow in
natural porous media”, Advances in Water Resources, (2016), under revision.
17. Alizadeh, A.H., and Piri, M., “The effect of saturation history on three-phase
relative permeability: An experimental study”, Water Resources Research,
(2014), 50, 1636-64.
18. Morrow, N. R., “Interfacial phenomena in petroleum recovery”, Surfactant
Science Series, V26, CRC Press, 1990.
19. Kestin, J., Khalifa, H. E., and Correia, R. J., “Tables of the dynamic and
kinematic viscosity of aqueous NaCl solutions in the temperature range 20-150°C
and the pressure range 0.1-35 MPa”, Journal of physical and chemical reference
data, (1981), 10, 1, 71-88.
20. Amin, R., and Smith, T. N., “Interfacial tension and spreading coefficient under
reservoir conditions”, Fluid phase equilibria, (1998), 142, 1, 231-241.

Table 1- Geometrical and petrophysical properties of the rock samples used in this study.
Sample Experiment D & L (mm) Porosity (%) Air permeability (mD)
A LSWF 4.6 & 52 28.32 -
B HSWF 4.64 & 51 28.97 -
C Dean-Stark 25.4 & 80 28.37 165
D XRD analysis End trim - -

Table 2- Physical properties of the fluids used in this study.

Viscosity at 60°C and 1000 Density at ambient condition
Fluid Experiment
psi (cP) (g/cc)
Low-salinity brine LSWF 0.3664* 0.999
High-salinity brine HSWF 0.3836* 1.029
Crude oil HSWF and LSWF 3.48 0.857
*Estimated from data available in the literature [18-20].
 SCA2016-003 11/12

Figure 1 - Porosity distributions along the length of Samples A and B obtained using micro-CT images.

Figure 2 - A schematic diagram of the core-flooding apparatus used in this study.

Figure 3 - Visualization of pore-scale fluid occupancies Figure 4 - Visualization of pore-scale fluid occupancies
generated at different stages of LSWF experiment generated at different stages of HSWF experiment
(resolution = 1.64 µ m; blue: water, red: oil, and gray: solid). (resolution = 1.64 µ m; blue: water, red: oil, and gray: solid).
 SCA2016-003 12/12

Figure 5 - Oil saturation distributions along the FOV at different stages of LSWF (left) and HSWF (right) tests.

Figure 6 - Histograms of in-situ contact angle distributions during different stages of LSWF (left) and HSWF (right) tests.

Figure 7 - Two-dimensional cross-sectional views of pore-scale fluid occupancy during primary LSWF test performed on a preserved
reservoir sandstone sample (resolution = 1.64 µ m; blue: water, red: oil, and gray: solid). The images show changes in wettability
during the LSWF and the subsequent invasion of water into an oil-filled pore.
 SCA2016-011 1/12

WETTABILITY EFFECTS ON OSMOSIS AS AN OIL

MOBILIZATION MECHANISM DURING LOW SALINITY
WATERFLOODING
S.B. Fredriksen1, A.U. Rognmo1, K. Sandengen2, and M.A. Fernø1
1
Department of Physics and Technology, University of Bergen, Norway
2
Statoil ASA, Rotvoll, Trondheim, Norway

This paper was prepared for presentation at the International Symposium of the Society of Core
Analysts held in Snowmass, Colorado, USA, 21-26 August 2016

ABSTRACT
This experimental study identifies osmosis as an oil mobilization mechanism during low
salinity waterflooding using capillary tubes and micromodels with different wetting
properties. Oil-wet capillary tube tests verify that oil acts as a semi-permeable membrane
for water transport when separating brines of different salinity. Osmotic oil mobilization
was investigated using two-dimensional silicon-wafer micromodels for direct optical
visualization of fluid interactions. Water diffusion was identified as the main transport
mechanism at strongly water-wet conditions and capillary equilibrium. Osmosis was
observed at both strongly water-wet and oil-wet conditions influenced by the pore-level
distribution of oil and water. At strongly water-wet conditions osmosis displaced oil by
expanding water-in-oil emulsions. At oil-wet conditions low salinity water was
transported by osmosis to otherwise inaccessible regions of high-saline brine. Osmotic oil
mobilization is a fluid-fluid interaction and should therefore be present in both sandstone
and carbonate reservoirs. The experimental results support this and provide evidence of
osmotic water transport and oil mobilization under various wettability conditions.

INTRODUCTION
Low salinity waterflooding (LSW) denotes injection of diluted brine concentrations in
either secondary or tertiary recovery mode to increase oil recovery. Whereas
conventional waterflooding uses formation brine or seawater predominantly to maintain
reservoir pressure, LSW improves microscopic sweep by injecting water with diluted salt
concentrations [1]. LSW has been reported to enhance oil production in a series of
coreflood experiments [2], and has been tested in various field operations, see e.g. [3].
There is a consensus that an optimal wetting preference exists at which LSW is favorable
for oil recovery. LSW is generally expected to shift reservoir wettability towards more
water-wet conditions.

The initial research on LSW emphasized on fluid-rock interactions and the mechanisms
that alter wettability of a rock surface. Coreflood results in sandstone provided a variety
of theories aiming at defining the low salinity effect (LSE), as summarized by Sheng [4].
One of the more accepted hypotheses is the concept of multi-component ion exchange
 SCA2016-011 2/12

(MIE), where inorganic Ca2+ and Mg2+ cationic exchange, between the negatively
charged sandstone and the injected low salinity water, replace organic polar components
at the rock surface to shift wettability towards water-wet [5]. The influence of cationic
exchange was partially confirmed by others, who also reasoned that a major contribution
was reduction of ionic brine strength [3]. Reduced ionic strength lowers the cationic
screening potential that result in electrostatic repulsion and expansion of the electrical
diffuse double layers surrounding clay and oil particles. Oil is desorbed from the surface
and the initial wetting state is altered [3].

When exposed to LSW, and contrary to the established hypotheses [3, 5], an increase in
recovery is also demonstrated in carbonates [6, 7]. Carbonate reservoirs are preferentially
oil-wet due to a net positively charged surface [8]. According to Ligthelm et al. [3],
reducing the ionic brine strength should yield greater attraction between organic polar
components and the positive carbonate surface causing stronger oil-wet conditions, not a
shift towards water-wet. The capability for multi-component cationic exchange is also
restricted considering the positively charged rock. The general understanding is therefore
that the underlying mechanism of LSE is yet to be found or that various mechanisms
exist for different crude-oil/brine/rock systems [4, 9, 10].

Osmosis had been largely overlooked as a plausible mechanism for improved oil
recovery during LSW [10]. Osmosis is a thermodynamic driving force for water transport
caused by a difference in chemical potential between two aqueous solutions separated by
a semi-permeable membrane [11]. In their initial work, Sandengen and Arntzen [10]
observed oil displacement when separating high –and low-saline solutions by an oil
membrane. A further investigation using oil-wet and water-wet micromodels showed that
the osmotic effect was influenced by water connectivity and it was believed that osmotic
gradients could not be established at water-wet conditions [12]. More recently, however,
it was found that osmosis may be active as a transport mechanism in water-wet systems
[13].

Current experimental data, obtained in micromodels, support the theory of osmosis as a

mechanism for pore-scale oil mobilization regardless of the crude-oil/brine/rock system.
Osmosis may therefore contribute to oil mobilization in both sandstone and carbonate
reservoirs. The experimental analysis herein confirm this hypothesis and compose a
collective theory identifying osmosis as one of the contributing mechanisms for oil
mobilization at pore-scale during LSW.

EXPERIMENTAL
Previous work on osmosis using water-wet capillary tubes could not clearly identify if
water transport occurred through the oil bulk or by film-flow induced by interfacial
tension [10]. In this study, oil-wet PTFE tubes are used to isolate and verify water
transport by osmosis through the oil bulk. In addition, silicon-wafer micromodels
replicating the two-dimensional pore-structure of natural sandstone are used to
qualitatively observe fluid-fluid interactions when performing wettability studies on
 SCA2016-011 3/12

osmotic oil mobilization during LSW. The strongly water-wet nature of the models is
ideal for investigating osmosis by isolating known processes that occur during LSW i.e.
wettability alteration towards more water-wet.

Fluids
Synthetic high-saline brines were made by adding given quantities of analytical-grade
chemicals to deionized water (DIW), while pure DIW was used as low-saline brine. Oils
were either toluene or crude oil (from a North Sea chalk reservoir). Toluene was used
because of high water solubility to accelerate flux during osmosis. Colorants such as Oil
Red O (red), Erinoglaucine disodium salt (blue) and Quinoline Yellow WS (yellow) were
added at a concentration of ~0.05 wt% to optically distinguish the fluids at pore-scale by
microscope. An overview of the fluids used and their characteristics are given in Table 1.

Table 1 – Chemical composition and characteristics of brines and oil phases.

Fluid ID Composition Ionic Osmotic Standard Interfacial
strength1) pressure2) molar tension
[M] [bar] entropy3) w/Oil A
[J/mol] [mN/m]
Brine A 5.0 M NaAc 5.0 124 138.2 (s) 14.0
Brine B 5.0 M NaCl 5.0 124 95.06 (aq) 25.3
Brine C 5.0 M CaCl2 12.5 124 123.87 (aq) 22.5
Brine D 1.0 M NaCl 1.0 25 95.06 (aq) 20.4
DIW Deionized water 0 0 69.95 (aq) N/A
Oil A (Toluene) C7H8 N/A N/A N/A N/A
Oil B (Crude) 53 wt.% HC N/A N/A N/A N/A
35 wt.% aromatics
12 wt.% resins
0.9 wt.% asphaltenes
HMDS >97% purity N/A N/A N/A N/A
1)
Assume complete dissociation [14].
2)
Assume non-electrolyte by Van’t hoff equation [14]. Provides the maximum osmotic pressure
that can develop in a solution when separated from deionized water.
3)
The entropy contribution from one mole of substance at standard conditions [15].

Oil-wet capillaries for verifying osmotic water transport

To verify osmotic water transport a series of capillary tube tests were performed using
oil-wet polytetrafluoroethylene (PTFE) tubes. The oil-wet nature restricts film flow along
the surface, and water transport should occur only through the oil-phase by the osmotic
gradient. Oil A was injected into the tube followed by high-saline Brine A and DIW on
either side. A baseline without an osmotic gradient was also established using two water
droplets of high-saline Brine A. The PTFE tubes were plugged and placed in an oven at
50 ºC for 20 days. Droplet movement was monitored and pictures were taken to
document the process.

Micromodels for studying wettability effects on osmotic oil mobilization

Silicon-wafer micromodels representing natural sandstone were used to investigate the
potential for osmotic oil mobilization by LSW at various wettability conditions. The
 SCA2016-011 4/12

strongly water-wet micromodels replicate a porous network in connection to adjacent

channels acting as fractures. One micromodel was prepared using high-saline Brine D as
connate water and Oil B as the oil-phase. At strongly water-wet conditions the high-
saline brine adhered to the pore walls and was continuous throughout the porous network.
Another system was prepared to represent oil-wet conditions. Here, Brine C was used as
the high-saline phase and Oil A as the oil-phase. The micromodels were rendered oil-wet
using Hexamethyl-diasilazane (HMDS) according to the modification procedure
described elsewhere [12].

When preparing the micromodels they were 100% saturated with Brine C or D as connate
water. Oil drainage was performed using either Oil A or B. After drainage, secondary
waterflooding was conducted using high-saline brine allowing spontaneous imbibition to
act as the main driving force until capillary equilibrium was reached. Baseline tests were
run prior to performing LSW through the adjacent channel establishing a salinity contrast
to the matrix high-saline brine. LSW was maintained at a flow rate of 1ml/day to
represent dynamic conditions through a fracture network with water transport to the
matrix dominated by small-scale diffusion processes.

A microscopic setup was used for pore-level visualization. Experiments were performed
at ambient temperature and pressure with the micromodels placed horizontally on the
microscope platform. Syringe pumps were connected to fluid distribution ports for brine
and oil injection. Field-of-view was carefully selected showing parts of the matrix and
adjacent flow channel. A camera and computer with associated imaging-software
connected to the microscope captured time-lapse-pictures for documentation and post-
image processing. LSW lasted between 7 to 14 days for each test.

RESULTS
Osmotic water transport in oil-wet capillary tubes
The effect of a salinity gradient on water transport through a semi-permeable oil
membrane using oil-wet capillary tubes is shown in Figure 1. The two image sequences
show water transport of low-saline water (yellow) through the oil-phase (red) into the
adjacent high-saline solution (white) over time. With a salinity gradient present (image
1), all of the low-saline water was transported through the oil-phase into the high-saline
brine after 20 days. The increased volume and color-shift indicates the direction of water
transport from the low salinity water to the high-saline brine. Without a salinity gradient
water transport was not observed (image 2), although some volume changes occurred due
to slow dissolution of water in toluene causing the high-saline droplets to shrink.
 SCA2016-011 5/12

Figure 1. Osmotic water transport in oil-wet capillary tubes. The yellow phase is low salinity water, the
red is the oil-phase and white is the high-saline brine. Time lapse images (1, 8, 11, 16 and 20 days) are
shown from top to bottom. 1) Osmotic gradients cause the low salinity water (yellow) to shrink as water is
transported through the oil phase into the high-saline solution. 2) The baseline result (i.e. high-saline
solution only) showed no osmotic water transport only a slight dissolution of water into toluene.

The interfacial tension (IFT) between oil and water varies with salinity, and decreases
upon dilution until reaching a minimum IFT at a critical salt concentration [16]. Hence,
when two solutions with different salinity are separated by an oil-phase, an uneven
balance is created across the oil-phase that may lead to convective water transport. IFT
was measured for a combination of brines with Oil A (cf. Table 1). To determine water
transport as an effect of osmosis only, combinations of Brine A, B and DIW were used in
capillary-tube tests. Brine A has a lower IFT compared to Brine B and behaves
qualitatively different in terms of capillary forces, but the two brines have the same ionic
strength and osmotic gradient (cf. Table 1).

Table 2 provides an overview of the capillary tube tests performed with configurations of
high – and low saline solutions separated by oil. Expansion of the reference solution is
defined by “+”, contraction by “-” and no change is termed “0”. In all cases the observed
results matched the direction of water transport predicted theoretically by osmosis.

Table 2 – Water transport in oil-wet capillary tube tests. The experimental results are correlated
with theoretical predictions from interfacial tension and osmosis.
Theoretical Theoretical
Reference Opposing Experimental
Prediction Prediction
solution solution Observation
Osmosis Interfacial tension
Brine B Brine B 0 0 0
Brine B Brine A 0 - 0
Brine B DIW + - +
DIW DIW 0 0 0
Brine A DIW + + +
Brine A Brine A 0 0 0

Wettability effects on osmotic oil mobilization

Oil mobilization during LSW occurs in response to forward osmosis when a system is
partially saturated with high-saline brine and oil [17]. Both strongly water-wet and oil-
wet silicon-wafer micromodels were used to evaluate the influence of wettability on
osmotic oil mobilization. To create a salinity contrast to the high-saline connate brine
 SCA2016-011 6/12

LSW was injected through the adjacent fracture only. No fluid displacement was
observed in baseline experiments, where the salinities of injected and connate brine were
equal.

Strongly water-wet conditions

At strongly water-wet conditions the initial high-saline brine adhered to the pore walls
creating a continuous water layer. Due to water connectivity and ceased capillary
imbibition the main transport mechanism was water diffusion from the low salinity phase
through film-flow diluting the high-saline connate brine. The increasing water-volumes
within the porous network resulted in a redistribution of fluids and capillary trapped oil
was viscously displaced into the fracture to give way for the preferred water phase.
Figure 2 shows how diffusion of low salinity water mobilized oil in a strongly water-wet
system.

Figure 2. Low salinity water diffusion and oil mobilization. Oil B (crude oil, brown), Brine D (blue),
DIW (turquoise) and rock grains (grey). Image 1-3 shows several pores, and Image 4 shows the fracture
located below (different length scale). 1) Initial fluid distribution. 2) Diffusion of DIW through pore A and
film-flow into pore B displacing oil to adjacent pores. 3) Continued DIW oil displacement from several
interconnected pores. 4) Fluid redistribution leads to oil displacement into adjacent fracture (yellow
arrows).

The osmotic effect was observed in small high-saline brine-droplets dispersed within the
oil-phase indicated by the yellow arrows in Figure 3, image 1. These water-in-oil
emulsions started to grow due to osmotic water transport during LSW and accumulated
into larger emulsions by coalescence and Ostwald ripening. This growth was not
observed without a salinity gradient (i.e. baseline). The continuous aggregation (image 2)
embodied most of the oil-phase, slowly displacing it towards the lower channel and into
connecting pore throats.
 SCA2016-011 7/12

Figure 3. Oil displacement by osmotic water-in-oil expansion. Oil B (crude oil, brown), Brine D (blue),
DIW (turquoise) and rock grains (grey). 1) The fluid distribution prior to performing LSW. An oil-phase is
located in the pore network in contact to the adjacent channel. Dissolved droplets of high-saline brine
within the oil are indicated by the yellow arrows. 2) During LSW, the water-in-oil emulsions start to grow
and the aggregation of emulsions displace the oil-phase.

Oil-wet conditions
At oil-wet conditions, osmotic gradients were established when the low salinity water
contacted the oil-phase leading to swelling of capillary trapped high-saline brine. Figure
4 shows expansion of capillary-trapped brine (blue) and subsequent oil (orange)
mobilization by osmosis. As the capillary trapped connate brine expanded the oil was
displaced towards the flow channel as indicated by the yellow arrows in image 2. The oil-
phase was mobilized into the fracture when the viscous pressure from brine expansion
exceeded the capillary pressure withholding oil in matrix (image 3).

Figure 4. Osmotic oil mobilization by capillary- trapped brine expansion. Oil A (orange), Brine C
(blue), DIW (turquoise) and rock grains (grey). 1) Fluid configuration after initiating LSW. Capillary
trapped regions of high-saline brine are found within the oil-phase. 2) The high-saline regions start to
expand in response to osmotic induced water transport mobilizing oil towards the fracture by the yellow
arrows. 3) Oil is produced from the porous medium and into the nearby flow channel.
 SCA2016-011 8/12

DISCUSSION
Verification of osmotic water transport in capillaries
The oil-wet capillary tube tests represent a situation where stagnant pockets of connate
water, either through snap-off or bypassed, are surrounded by oil. The direction of water
transport observed in Figure 1 matched prediction by osmosis, and suggests limited
influence of salinity effects on IFT. This is also corroborated by previous qualitative
observations in water-wet micromodels [10]. Rate of oil mobilization by osmosis
depends, among other factors, on the thickness of the oil-phase. It can be compared to the
diffusion rate by Fick’s law, given the diffusion coefficient by the Stokes-Einstein (SE)
equation. SE assumes that water diffusion is inversely proportional to the oil-phase
viscosity. However, when water moves through an oil-phase without hydrogen bonding
the smaller water molecules diffuse faster than predicted [18], and additional
considerations are required to describe the waterflux.

The pore-scale osmotic effect observed in capillary tubes is relevant during spontaneous
imbibition, where oil production is determined by capillary forces without viscous flow.
It is believed that the osmotic effect may contribute favorably during core-scale
spontaneous imbibition of low salinity water by expanding volumes of high-saline brine
and mobilizing oil out of the core.

Verification of water diffusion in micromodels

Water diffusion was observed as the main transport mechanism at strongly water-wet
conditions by the connectivity between fracture bulk-flow and matrix brine. The transport
is induced by the chemical potential gradient between the high- and low-saline phases.
The potential is defined as the sum of Gibbs free energy for all components in the
solution. When salts are mixed with water to form high-saline brine the components
spontaneously dissolve into electrically charged ions due to the resulting increase in
entropy [11] (cf. Table 1). Heat is released during the dissolution process and the Gibbs
free energy of the solution, e.g. the energy available to do work, is reduced. Low salinity
water (DIW in this work) has a higher chemical potential than high-saline brine, and
water will move from regions of high to low potential as observed in Figure 2. The
system is stabilized when Gibbs free energy reaches a minimum for the system as a
whole. Hence, water will diffuse to re-establish thermodynamic stability.

Oil mobilization by water diffusion

It was mainly the diffusion of solvents (i.e. water) that diluted the high-saline brine in
matrix to equilibrate the difference in chemical potential. Ion-dipole forces between salt
ions and water molecules reduce the diffusion of solutes as they are stronger than the
induced hydrogen bonds between water molecules [19]. Water diffusion was initially
identified as a dilution mechanism causing film-expansion and droplet growth during
LSW [13]. It was reasoned that the accumulation of water in local water-films increased
the water pressure leading to viscous oil displacement into nearby flow channels.
Research by Stoll et al. [20] emphasized that in absence of significant viscous and
capillary forces, wettability modifiers must initially diffuse under its own concentration
 SCA2016-011 9/12

gradient into the matrix to activate its chemical process. The results in Figure 2 support
these findings and illustrate how propagation of water may re-distribute the fluids in
matrix, displacing oil from pore-to-pore and towards the fracture.

Verification of osmotic water transport in micromodels

Oil mobilization was observed at both strongly water-wet and oil-wet conditions, with
water transport to otherwise inaccessible regions of high-saline brine and water-in-oil
emulsions (Figure 3 and 4). Osmosis is the transport of water when high- and low-saline
brines are separated by a semi-permeable membrane that is permeable to pure water, but
impermeable to ions [11]. To balance the chemical potential gradient, e.g. osmotic
gradient, the solvent must have equal values in the two brines by having the two brines at
different pressures [14]. Hence, water is transported from the low-saline brine through the
membrane to compensate for the reduction in chemical potential of the high-saline brine.
The osmotic pressure within the high-saline brine increase with addition of water
molecules, and transport stops when the osmotic pressure balances the flow induced by
the osmotic gradient.

Semi-permeable membrane characteristics of oil

The semi-permeable properties of oil depend on the intermolecular forces acting between
the solute, solvent and hydrocarbon compounds present. The capacity for oils to bind
with hydrogen depends on the hydrocarbon composition. Mineral oils with chains of
alkanes have zero dipole moment and hence no capacity to hydrogen bond [18].
Asphaltenes and some aromatics present in most crude oils have structures to create weak
hydrogen couplings that influence the flux of water [21, 22]. Hence, an oil-phase with a
higher fraction of non-polar alkanes should have better membrane properties and act as a
more efficient conduit for water molecules under the influence of osmotic gradients.

Wettability effects on osmotic oil mobilization

At strongly water-wet conditions, isolated high-saline brine existed only as water-in-oil
emulsions that expand during LSW by aggregating inside the phase. Researchers have
drawn special attention to the growth of water-in-oil emulsions: some explain it as a
result of surface-active components arranging micelles transporting water molecules to
inaccessible brine, see e.g. [23], whereas others have suggested that the water transport is
caused by osmosis [15-13]. In addition, it has been shown that water-in-oil emulsions can
grow without the presence of surface-active components by using mineral oil during
LSW [15]. It is therefore suggested that the growth of water-in-oil emulsions is a result of
osmosis, supported by results from the present (Figure 3 and 4) and previous work [13].

Water-in-oil emulsions are present in this work independent on wettability. The stability
of water-in-oil emulsions depend on the composition of the oil-phase. Oil B, i.e. crude
oil, contains asphaltenes that help stabilize emulsions [24]. Mineral oils containing non-
polar alkanes, as used in the oil-wet system, are not interfacially active and cannot form
stabilized films around the water-droplets at the oil/water interface [24]. As long as there
are compounds present that can form stable films, emulsions should develop.
 SCA2016-011 10/12

An oil-wet system represents the most ideal for osmosis as it holds a number of suitable
membranes and confined volumes of high-saline brine for oil mobilization (Figure 4).
Several isolated regions of high-saline brine started to expand and the fluids within
matrix were redistributed as observed in images 1-2. As a result oil was mobilized into
the nearby flow-channel (Figure 4, image 3).

The calculations of osmotic pressure in Table 1 indicate a high potential for viscous
pressure build-up during LSW. These theoretical values are unlikely to develop within
the pore-space due to the imperfect membrane properties of the oil. In corefloods the
reported pressure response is usually transient or modest at best when injecting low
salinity water [1, 7, 10, 25]. If osmosis is a significant contributor to the overall trend one
would observe a general increase in pressure followed by a slow decrease as the gradient
is gradually balanced, see e.g. [25]. However, a relationship cannot be drawn between the
pressure gradients reported in literature and osmosis.

Osmosis as low salinity effect on reservoir scale

LSE is more relevant at oil-wet conditions as the general understanding is that LSW
shifts wettability towards more water-wet increasing oil recovery. However, it has been
documented that LSE improves oil recovery without the necessary conditions, see e.g. [1,
25]. The variety of circumstances under which LSE may or may not be observed suggests
that more than one mechanism may be in play [1]. In the present work, osmosis is
observed independent on oil/brine/rock interactions and should therefore be applicable to
both oil and water-wet systems.

In a highly flooded water or mixed-wet reservoir it is suggested that enclosed volumes

and emulsions of connate water will swell in response to osmosis. Re-mobilization of oil
will occur at pore-level until the oil-films either ruptures or thermodynamic stability is re-
established. However, it is not likely that the effect will greatly reflect on total recovery.
A fractured, oil-wet reservoir (i.e. Figure 4) would be an ideal system for osmosis to
have a significant impact on oil recovery. In a fractured system osmosis will act as a
driving force for water transport into the oil-wet matrix. Osmosis may contribute
positively to open pathways that increase the water relative permeability, aiding other
effective low salinity mechanisms to take place, such as wettability alteration in
otherwise inaccessible regions of the matrix, and for oil to be produced into high
permeable flood zones.

As osmosis occur independent on wettability is likely to be present in both sandstone

(water-wet) and carbonate (oil-wet) formations. To properly determine what effect
osmosis will have on recovery, one could use model systems where wettability alteration
is excluded. This could help to better quantify the extent of osmosis during LSW. The
present work identifies osmosis as a pore-level LSE that improves microscopic sweep in
combination with other low salinity mechanisms.
 SCA2016-011 11/12

CONCLUSIONS
• Oil-wet capillary tube tests verified osmotic water transport as oil-bulk diffusion
from a solution of low salinity to a solution of high salinity due to the semi-
permeable characteristics of oil.
• The work performed substantiate osmosis as an underlying low salinity effect for
oil mobilization at pore-level by identifying osmotic water transport at both
strongly water-wet and oil-wet conditions during low salinity waterflooding.
• Osmotic water transport mobilized oil by either swelling water-in-oil emulsions or
expanding otherwise inaccessible regions of capillary trapped high-saline brine.
• At strongly water-wet conditions and capillary equilibrium water diffusion by
film-flow was the main transport mechanism for oil mobilization.
• The osmotic effect is influenced by fluid distribution of oil and water. It is
independent of system wettability and should therefore be present as a pore-scale
oil mobilization mechanism in both sandstones (e.g. water-wet) and carbonates
(e.g. oil-wet) during LSW.

ACKNOWLEDGEMENTS
The authors would like to thank the Norwegian Research Council for financial support.

REFERENCES
[1]. Morrow, N. and Buckley, J. “Improved Oil Recovery by Low-Salinity Waterflooding,”
Journal of Petroleum Technology, (2011) 63, 05, 106-112.
[2]. Tang, G. and Morrow, N. “Salinity, Temperature, Oil Composition, and Oil Recovery by
Waterflooding,” SPE Reservoir Engineering, (1997) 12, 04, p. 269-276.
[3]. Lightelm, D., Gronsveld, J., Hofman, J., Brussee, N., Marcelis, F. and van der Linde, H.
“Novel Waterflooding Strategy by Manipulation of Injection Brine Composition,” Society of
Petroleum Engineers, (2009).
[4]. Sheng, J. “Critical review of low-salinity waterflooding,” Journal of Petroleum Science and
Engineering, (2014) 120, p. 216-224.
[5]. Lager, A., Webb, K., Black, C., Singleton, M. and Sorbie, K. “Low Salinity Oil Recovery –
An Experimental Investigation1,” Society of Petrophysicists and Well-Log Analysis, (2008) 49,
01, p. 28-35.
[6]. Romanuka, J., Hofman, J., Ligthelm, D., Suijkerbuijk, B., Marcelis, A., Oedai, S., Brussee,
N., van der Linde, H., Aksulu, H. and Austad, T. “Low Salinity EOR in Carbonates,” Society of
Petroleum Engineers (2012).
[7]. Yousef, A., Al-Saleh, S., Al-Kaabi, A. and Al-Jawfi, M. “Laboratory Investigation of the
Impact of Injection-Water Salinity and Ionic Content on Oil Recovery From Carbonate
Reservoirs,” SPE Reservoir Evaluation & Engineering, (2011) 14, 05, p. 578-593.
[8]. Chilingar, G. and Yen, T. “Some Notes on Wettability and Relative Permeabilities of
Carbonate Reservoir Rocks, II,” Energy Sources (1983) 7, 01, p. 67-75.
[9]. Suijkerbuijk, B., Hofman, J., Ligthelm, D., Romanuka, J., Brussee, N., van der Linde, H. and
Marcelis, A. “Fundamental investigations into wettability and low salinity flooding by parameter
isolation,” Society of Petroleum Engineers (2012).
[10]. Sandengen, K. and Arntzen, O. “Osmosis During Low Salinity Water Flooding,” EAGE 17th
European Symposium on Improved Oil Recovery (2013).
 SCA2016-011 12/12

[11]. Allen, J. Biophysical Chemistry, Wiley-Blackwell, Singapore (2009) p. 47-51, 54-56, 73,
94-95, 120-123.
[12]. Sandengen, K., Kristoffersen, A. and Melhuus, K. “Osmosis as Mechanism for Low-
Salinity Enhanced Oil Recovery,” SPE Journal, (2016) p.1-9.
[13]. Fredriksen, S., Rognmo, A. and Fernø, M. “Pore-Scale Mechanisms during Low Salinity
Waterflooding: Water Diffusion and Osmosis for Oil Mobilization,” Society of Petroleum
Engineers, (2016).
[14]. Mortimer, R. Physical Chemistry, Academic Press, United States of America (2000), p. 168-
170, p. 248-249.
[15]. The National Institute of Standards and Technology (NIST). “NIST Standard Reference
Database 69; NIST Chemistry WebBook,” U.S. Secretary of Commerce, United States of America
(2016).
[16]. Moeini, F., Hemmati-Sarapardeh, A., Ghazanfari, M., Masihi, M. and Ayatollahi, S.
“Toward mechanistic understanding of heavy crude oil/brine interfacial tension: The roles of
salinity, temperature and pressure,” Fluid Phase Equilibira (2014) 375, 01, p. 191-200.
[17]. Cath, T., Childress, A. and Elimelech, M. “Forward osmosis: Principles, applications, and
recent developments,” Journal of Membrane Science (2006) 281, 1-2, p. 70-87.
[18]. Su, J., Duncan, B., Mornaya, A., Jutila, A. and Needham, D. “The effect of hydrogen
bonding on the diffusion of water in n-alkanes and n-alcohols measured with a novel single
microdroplet method,” The Journal of Chemical Physics, (2010) 132, 04, p. 0445061-8.
[19]. Person, E., Golden, D. and Royce, B. “Salting Effects as an Illustration of the Relative
Strength of Intermolecular Forces,” Journal of Chemical Education (2010) 87, 12, p. 1332-1335.
[20]. Stoll, W., Hofman, J., Ligthelm, D., Faber, M. and van den Hoek, P. “Toward Field-Scale
Wettability Modification – The Limitations of Diffusive Transport,” SPE Reservoir Evaluation &
Engineering (2008) 11, 03 p. 633-640.
[21]. Spiecker, P., Gawrys, K. and Kilpatrick, P. “Aggregation and solubility behavior of
asphaltenes and their subfractions,” Journal of Colloid and Interface Science (2003) 267, 01, p.
178-193.
[22]. Levitt, M. and Perutz, M. “Aromatic Rings Act as Hydrogen Bond Acceptors,” Journal of
Molecular Biology (1988) 201, 04, p. 751-754.
[23]. Emadi, A. and Sohrabi, M. “Visual Investigation of Oil Recovery by Low Salinity Water
Injection: Formation of Water Micro-Dispersions and Wettability Alteration,” Society of
Petroleum Engineers (2013).
[24]. Kokal, S. “Crude-Oil Emulsions: A State-Of-The-Art Review,” SPE Production and
Facilities (2005) 20, 01, p. 5-13.
[25]. Cissokho, M., Boussour, S., Cordier, P., Bertin, H. and Hamon, G. “Low Salinity Oil
Recovery on Clayey Sandstone: Experimental Study,” Society of Core Analysists (2009).
 SCA2016-017 1/12

Micro-CT study of the Impact of Low Salinity Waterflooding

on the pore-scale fluid distribution during flow
W.-B. Bartels1, 2, M. Rücker1, 3, S. Berg1, H. Mahani1, A. Georgiadis1, 8, N. Brussee1, A.
Coorn1, H. van der Linde1, A. Fadili1, C. Hinz4, A. Jacob4, C. Wagner5, S. Henkel6, F.
Enzmann4, A. Bonnin7, M. Stampanoni7, H. Ott3, 9, M. Blunt3, S.M. Hassanizadeh2
1
Shell Global Solutions International B.V., Kesslerpark 1, 2288 GS Rijswijk, The Netherlands
2
Earth Sciences department, Utrecht University, 3584 CD Utrecht, The Netherlands
3
Department of Earth Science and Engineering, Imperial College London, SW7 2AZ UK
4
Geosciences Institute, Johannes Gutenberg-University, Becherweg 21, 55099 Mainz, Germany
5
Math2Market GmbH, Stiftsplatz 5, 67655 Kaiserslautern, Germany
6
Institute of Geosciences, Friedrich Schiller University Jena, Burgweg 11, 07743 Jena, Germany
7
Swiss Light Source, Paul Scherrer Institute, CH-5232 Villigen, Switzerland
8
Department of Chemical Engineering, Imperial College London, SW7 2AZ UK
9
Department of Petroleum Engineering, Montanuniversität Leoben, A-8700 Leoben, Austria

This paper was prepared for presentation at the International Symposium of the Society of Core
Analysts held in Snowmass, Colorado, USA, 21-26 August 2016.

ABSTRACT
Many studies indicate that the recovery of crude oil by waterflooding can be improved in
both sandstone and carbonate rocks by lowering the salinity of injected water. This so-
called low salinity effect is thought to be associated with the change of the wetting state
of rock towards more water wet. However, it is not very well understood how wettability
alteration on the pore level could lead to an increase in production at the Darcy scale.
Therefore, this study aims at direct pore scale observation of the wettability-change-
driven fluid reconfiguration related to a low salinity (LS) flood at the length scale
between a single pore and the Darcy scale (i.e. pore network scale). We investigate the
low salinity effect in real time and in 3D using synchrotron beamline-based fast X-ray
computed tomography during flow experiments.
Cylindrical outcrop rock samples of 20 mm length and 4 mm diameter were initialized by
first saturating them with high salinity (HS) brine and then with crude oil. Subsequently,
they were aged at 30 bars and 70°C for one week in order to establish wettability states
assumed close to reservoir conditions. The synchrotron beamline-based fast tomography
allowed us to image the pore scale fluid distribution at a spatial resolution of 3 µm and
(under flowing conditions) at time intervals of 10 s for a full 3D image.
The micro-CT flow experiments were conducted on both sandstone and carbonate rocks,
all in tertiary mode, i.e. by first performing a HS water flood i.e. forced imbibition (as
base-line) followed by a LS waterflood, both at identical field relevant flow rates. The
real-time imaging shows a saturation change during the HS waterflood which approaches
a stable equilibrium at its end. When switching to low salinity water, we observe a
 SCA2016-017 2/12

change in average saturation and pore scale distribution of both fluids, which is distinctly
different from the stabilizing saturation during HS flooding. Compared to the end of the
HS flood, during the LS flood in both sandstone and carbonate rock, the oil moves from
pore throats to the center of pore bodies. This movement is indicative of a pore scale
wettability transition from a mixed wet to a more water wet state. This process involves
(re)connection and disconnection of the oil phase as it moves through narrow pore throats
which is characteristic of ganglion dynamics.

INTRODUCTION
Low salinity waterflooding (LSF) is an enhanced oil recovery/improved oil recovery
(EOR/IOR) technique which mobilizes more oil by modifying and/or lowering the ionic
composition of the injected brine. For sandstones, the range at which additional
production occurs lies between 1500 and 5000 ppm. Lower salinities may lead to
formation damage due to fines migration and plugging. For carbonates, the LS threshold
at which the low salinity effect (LSE) occurs lies much higher. There are even cases
reported for sea water; see [1] and references therein for a short overview. In addition, the
exact brine chemistry is deemed more important for carbonates than for sandstones. The
LSE is often described phenomenologically by wettability alteration to a more water-wet
state. However, the mechanisms behind the alteration are not fully understood.
Several mechanisms have been proposed in the literature; however, there is no general
consensus on the dominant mechanism or mechanisms; see e.g. [2-5]. Moreover, the
evidence supporting a particular mechanism is often indirect or inferred from the
experimental measurements; and in some cases the evidence is even contradictory. If we
examine the body of literature, in general it can be divided into two groups. On the one
hand there are studies looking at contact angle changes and detachment of crude oil from
mineral surfaces i.e. at sub pore scale. On the other hand there are experiments which
look for incremental production of crude oil in core flooding experiments. However, it is
not clear how a contact angle change on the sub pore scale leads to additional Darcy scale
oil production. There is a gap between the (sub) pore scale and the Darcy scale in which
hardly any research has been conducted. Incremental oil production on the Darcy scale
requires a change in configuration of oil on the scale of multiple pores, i.e. the pore
network scale. The details of this change of oil saturation distribution on the pore
network scale is expected to close the gap between surface science and reservoir
engineering concepts and provide the links between proposed mechanisms and Darcy
scale observation.
In order to investigate the effect of LSF on the configuration of oil on this intermediate
pore network scale under flowing conditions, we use synchrotron based fast X-ray
tomography.

MATERIALS AND METHODS

Rock samples and brines
For the experiments in this study, we used Berea sandstone [6] combined with reservoir
dead crude “S” (density ρ = 0.87 g/cm3, viscosity η = 7.87 mPa·s, at T = 20°C) and
 SCA2016-017 3/12

Ketton as carbonate rock with reservoir dead crude “C” (ρ = 0.83 g/cm3, η = 4.30 mPa·s,
at T = 20°C). The samples were 20 mm long and 4 mm in diameter and embedded in
heat-shrunken PEEK (Polyether Ether Ketone) tubes. As HS brine, we used 200 g/L KI
(potassium iodide, ionic strength 0.602 mol/L) and as LS water, 27 g/L KI (ionic strength
0.081 mol/L).
Experimental workflow
A rock sample was first cleaned using isopropanol and then shrunken in PEEK. When
handling the core, nitrile gloves were used at all times to prevent altering the wettability
of the rock upon touch. The sample was then fully saturated with HS brine under
vacuum. Subsequently, it was mounted on the flow cell as used in [7, 8] to establish
initial water and oil saturation in the sample. When the sample was saturated with HS
brine and crude oil, the initial wettability state was established by aging it submerged in
crude oil at 30 bars and 70°C for one week. The experimental set-up we used has been
described in [7, 8], where the two cylinders of the in-situ pump were used for fractional
flow experiments. In this study, the cylinders were filled with HS brine and LS water,
respectively, to conduct tertiary LS flooding experiments.
The flooding and X-ray saturation monitoring experiments were performed at the
TOMCAT beamline of the Swiss Light Source (Paul Scherrer Institute) which is a fast
synchrotron-based X-ray computed micro-tomography facility. In this study, a “pink
beam” configuration with a finite X-ray energy spectrum was used. This setting caused
significant beam hardening. The flow rate in both high and low salinity experiments was
30 µl/min (equivalent to ~12 ft/day) and injection always occurred from bottom to top.
The field of view (FOV) was chosen 2 mm above the inlet, and dimensions of the
monitored area were approximately 4 mm laterally and 3 mm along the principal axis of
the sample. A full 3D image with a voxel size of 3 µm was obtained within about 7 s. The
transition from HS flooding to LS flooding was monitored continuously for 5 min. In
addition, images from 30 minutes before to 30 minutes after the transition were recorded
at a rate of 1 image/min.

Image processing
The images were reconstructed using the Paganin method [9, 10] since the end result was
found superior to that of the attenuation contrast reconstruction method, as shown in
Figure 1.The reconstructed micro-CT images were filtered, segmented, and processed
with the software package AVIZO 9.0 and 8.1 (FEI). A combination of gradient images
and watershed segmentation led to satisfactory results. After segmentation, the binarized
images are analyzed further.

RESULTS AND DISCUSSION

Potential Imaging Artifacts
The processed grey scale images before segmentation, such as the one for the oolitic
carbonate rock shown in Figure 2, show a clear contrast difference between grains, the
doped brine (at HS), and a dark phase which should contain the oil. However, a more
 SCA2016-017 4/12

detailed inspection reveals a grey- scale variation and a structure inside the darkest phase,
see
Figure 3, which occurs systematically for all samples studied. That raises the suspicion
that a third fluid phase may be present and brings up several questions about

Figure 1: Difference in image quality by reconstructing Berea sandstone with Paganin (left) and Filtered
Back Projection (right) algorithms. There is a clear difference in noise level and contrast between the two
methods. The scale bar indicates 200 micrometer.

significance and interpretation of the images and the origin of the phases that are visible.
This is discussed in the following sections.
Possible explanations for the apparent presence of three phases are:
A) artefacts related to reconstruction, filtering and beam hardening,
B) trapping of gas during sample preparation,
C) gas formation due to X-ray exposure,
D) degassing of crude oil due to X-ray exposure,
E) discoloration of a part of the brine phase due brine-rock interaction such as ion
exchange and adsorption
F) formation of oil/water structures.
Some of these possibilities are more likely than others. The questions that arise are: is this
an artifact that invalidates the experiment as such (items B-D) or an artifact that
complicates the data analysis (A); are the experiments in principle valid; or are we
actually dealing with a relevant phenomenon (E-F).
Note that we have observed these effects in all our synchrotron beamline LSF
experiments in both sandstone and carbonate samples. Explanations B, C and D relate to
the origin of the darkest phase, suggesting that the intermediate phase is oil and the bright
phase is HS brine. Explanations D and E assume that the intermediate phase is the
anomalous phase i.e. something other than oil or brine. Below, we provide a detailed
assessment of each explanation.

A) Artefacts related to reconstruction, filtering and beam hardening

 SCA2016-017 5/12

Our data shows significant beam hardening. To what extent this effect translates into gray
scale variations and structures within pores is not clear. In addition, glow from the HS
brine phase and from the grains may elevate the gray levels of the darkest phase in the
pore space. Still, the darkest phase seems to be much brighter than the exterior, indicating
that it cannot be a gas phase.

Figure 2: Different gray values in the pore space of oolitic carbonate rock. We distinguish between bright,
intermediate and dark phases. In some cases, interfaces are visible, see dashed arrows.

Figure 3: Stepped change in gray value and gradual changes in gray value overlap making segmentation
difficult. This example was taken from carbonate rock.

B) Trapping of gas during sample preparation

After the preparation of the samples at Shell laboratories, micro-CT scans with a
benchtop instrument (ZEISS Xradia 520 Versa) were made. Benchtop machines use a
much broader and higher energy X-ray spectrum but at several orders of magnitude lower
intensity. In those scans, no such variation in the darkest phase was observed; at least, the
formation of a separate gas phase could be ruled out. After preparation of the samples, in-
house micro-CT scans only revealed two sets of gray levels: that of HS brine (bright) and
oil. In cases that air and/or gas had been accidentally trapped in the pore space, this was
 SCA2016-017 6/12

clearly recognizable in the in-house scans and these samples had been excluded from
further analysis. However, imaging at the synchrotron beamline – approximately after
one week of aging –revealed the situation shown in Figure 2 in all samples. So, unless a
gas phase forms (C, D) upon exposure to the much higher intensity synchrotron beam,
explanation (B) can be ruled out.

C) Gas formation due to X-ray exposure

In previous beamline experiments, the formation of bubbles has been observed, but it was
also found that occurrence and magnitude of the effect depended on beam energy,
intensity, type of fluid, and dopant concentration. While the reason was ultimately not
clear, a photochemical effect was discussed as the most likely cause. In order to assess
whether the darkest phase in Figure 2 could be in-situ formed gas bubbles, we conducted
a long-time exposure experiment to see if the specific beam settings used in this study
would lead to bubble formation as well. Similarly to previous studies, we observed
bubble formation in the (doped) brine phase, as shown in Figure 4, but no formation of
gas bubbles in the oil phase were observed. Morphology and growth kinetics of the
bubbles are very different from the structure in the darkest phase in Figure 2. In addition,
if the bubbles have a photochemical origin, one would expect the process to occur or
intensify when freshly doped HS brine was injected in the system. Examining saturation
of the dark, intermediate and bright phase versus time displayed in Figure 7 does not
show such an increase. Therefore, (C) is also very unlikely.

Figure 4: Bubble formation in sandstone only filled with strongly doped HS brine. There was no flow
during this experiment. The images are taken with intervals of 10 seconds.

D) Degassing of crude oil

When the temperature is increased or the pressure is decreased, light hydrocarbon
components present in the crude oil may come out of solution and form gas bubbles.
However, the temperature during transport and experimentation did not increase by more
than 2 °C. In addition, the pressure was constant. Given that both crude oils were “dead
crude” (i.e. degassed crude) with the bubble point being far away from conditions in the
experiment, (D) is very unlikely.

E) Discoloration of part of the brine phase

 SCA2016-017 7/12

Some of the intermediate grey levels could represent brine with a reduced concentration
of contrast agent KI. One might speculate that such a compositional change might have
occurred through brine-rock interactions, such as ion exchange reactions with the solid,
or adsorption. However, the laws of thermodynamics would prohibit any spontaneous de-
mixing. Another counter-argument is that when the intermediate grey level is classified
as a brine phase, it implies very low initial oil saturation. In addition, if the intermediate
grey levels are interpreted as a separate phase, then the observation of a “contact angle”
with the brine phase would rule out an aqueous phase even though the phase is mostly
located in the corners of pores.

F) Formation of oil/water structures

Oil-brine mixing tests revealed that the crude oils used in this study can form oil/water
structures, such as emulsions, when in contact with brine. This is shown in Figure 5.

Figure 5: Different degrees of oil/water structure formation for crude “S” and “C”. The right hand side
shows a benchtop micro-CT scan of the interface between crude “C” and emulsion. The encircled portion
of this image shows droplets below the imaging resolution.

Such structures have also been described by others [11-13]. The gray values of these
oil/water structures would lay in between that of the pure phases. Any micro-structure i.e.
individual droplets may be significantly blurred given the limited spatial resolution of the
micro-CT and imaging artifacts discussed in (A).

In summary, the development or presence of a gas phase (explanations B-D) and the
presence of a second brine phase (E) seem unlikely. Certainly, imaging and
reconstruction artifacts occur (A) and if these are solely responsible for the variation of
grey levels, all should be lumped into the oil phase. But the formation of oil/water
structures (F) could occur as observed in tube tests under similar conditions (see Figure
5). Because of the potential connection of gray-scale feature to a LSE [12, 13] it might
therefore be beneficial to treat it until further clarification as a separate phase.
In the following we will discuss the response to LS and focus in particular on the aspects
that are robust against the considerations A-F. Therefore we will focus on the behavior of
the darkest phase.
 SCA2016-017 8/12

Response to High Salinity

During HS flooding, we see the oil phase clusters disconnecting and reconnecting, which
indicates a ganglion dynamics flow regime as observed already in 2D micro-models [14]
and other micro-CT imbibition experiments [8], [15], [16]. However, we see oil-filling
events covering pore bodies in the mixed-wet condition as shown in Figure 6. We did not
observe this in previous studies under water-wet conditions.

Figure 6: 3D rendering of pore filling event in carbonate (Ketton) rock in HS.

Figure 7: Saturation versus time plots for a HS flood in sandstone and carbonate. The decrease of the
darkest phase suggests that it cannot be gas that is created because of X-ray exposure. The initial
 SCA2016-017 9/12

saturations suggest that the intermediate gray phase and the dark phase should be considered as the same
phase.

Figure 8: When changing from HS to LS, a clear redistribution of oil can be seen from the pore throats to
the pore bodies. This is seen most clearly in Ketton rock (Carbonate).

Response to Low Salinity

The first observation in the images captured during LS is that we can track the change in
salinity as the concentration of contrast agent decreases as seen in Figure 8. Secondly,
we see a change in oil configuration as the oil phase moves from smaller to larger pores
at the transition from HS to LS, also shown in Figure 8. This is indicative of a wettability
alteration towards more water wet, as is expected in a successful LSF experiment. For
sandstone, this is intriguing since the salinity level used here is much higher than the
generally accepted salinity range of 1500-5000 ppm in which the macroscopic (Darcy)
LSE has been typically observed.

Pore filling events

We also observe ganglion dynamics during the LS flood. During LS flooding, we clearly
see that the extent of oil displacement is much larger than at the final stages of the HS
flood, as illustrated in Figure 9. This indicates that the LS water did mobilize more oil,
which may have come from the upstream part of the sample outside the FOV. This oil
saturation increase in the FOV can be temporary (a transient effect) and could be related
to oil banking by LS. The oil bank could eventually be produced from the sample.
 SCA2016-017 10/12

Figure 9: Top: Increase in the darkest phase in 7 seconds in Berea sandstone during LS flooding. This
magnitude of increase cannot be caused by gas. In addition, this increase is occurring after the saturation of
all phases stabilized during HS flooding, see Figure 7. Bottom: Cluster dynamics during LS flow in Berea
sandstone. Scale bar indicates 100 micrometer. The circle indicates growth of the oil cluster and the arrow
indicates a (re)connection point. Flow is from bottom to top.

CONCLUSIONS
We have observed an additional oil/water structure or a third phase and are not sure of its
nature and origin. Further tests need to be conducted to verify whether this oil/water
structure is a (micro-) emulsion. Regardless of the occurrence of this structure, we still
can draw the following conclusions regarding the LSF experiments.

• We reached a stable saturation during HS flooding.

• During LS injection, we observed the following:
i) There is oil redistribution from pore throats to pore bodies, which would
correspond to a shift of wettability to a more water wet system.

ii) There is an increase in the overall oil saturation in the investigated FOV. This is
most likely the mobilized or re-connected oil from upstream and is indicative of a
low-salinity effect. The increase of oil saturation can be transient and may be
 SCA2016-017 11/12

related to oil banking by LS. The oil bank, because of high saturation and high
relative permeability, may eventually be produced from the sample.

Additionally:
• For sandstones, the LSE occurred at relatively high salinities.
• We observed filling events covering multiple pores in both rock types, which we
believe to be specific to the mixed wet state of the system.
• We see ganglion dynamics in both HS and LS regimes but in LS we also see
remobilization of crude oil.

ACKNOWLEDGEMENTS
We would like to acknowledge the staff of the TOMCAT beamline at the Swiss Light
Source of the Paul Scherrer Institute, Villigen, Switzerland for support during the
execution of this work. We thank Alex Schwing and Rob Neiteler for the design of the
flow set-up and instrumentation, Fons Marcelis for sample preparation, Kamaljit Singh
for supplying the Ketton rock and Axel Makurat for helpful discussions and leadership
support. We gratefully acknowledge Shell Global Solutions International B.V. for
permission to publish this work.

REFERENCES
1. Mahani, H., A. Keya, S. Berg, W.-B. Bartels, R. Nasralla and W. Rossen, “Insights
into the Mechanism of Wettability Alteration by Low-Salinity Flooding (LSF) in
Carbonates”, Energy and Fuels (2015), 29, 1352-1367.
2. Aladasani, A., B. Bai, Y.-S. Wu and S. Salehi, “Studying low-salinity waterflooding
recovery effects in sandstone reservoirs”, Journal of Petroleum Science and
Engineering (2014), 120, 39-51.
3. Morrow, N., and J. Buckley, “Improved Oil Recovery by Low-Salinity
Waterflooding”, Journal of Petroleum Technology, SPE-129421 (2011).
4. Myint, P., and A. Firoozabadi, "Thin liquid films in improved oil recovery from low-
salinity brine." Current Opinion in Colloid & Interface Science (2015) 20, 2, 105-
114.
5. Lager, A., K. Webb, C. Black, M. Singleton and K. Sorbie, “Low Salinity Oil
Recovery – An Experimental Investigation”, Petrophysics (2008), 49, 1, 28-35.
6. Berg, S., H. Ott, S.A. Klapp, A. Schwing, R. Neiteler, N. Brussee, A. Makurat, L. Leu,
F. Enzmann, J.O. Schwarz, M. Kersten, S. Irvine and M. Stampanoni, “Real-time 3D
imaging of Haines jumps in porous media flow”, Proceedings of the National
Academy of Sciences (2013), 110, 10, 3755-3759.
7. Armstrong, R., H. Ott, A. Georgiadis, M. Rücker, A. Schwing and S. Berg,
“Subsecond pore-scale displacement processes and relaxation dynamics in multiphase
flow”, Water Resources Research (2014), 50, 9162–9176.
 SCA2016-017 12/12

8. Rücker, M., S. Berg, R. Armstrong, A. Georgiadis, H. Ott, L. Simon, F. Enzmann, M.

Kersten and S. de With, “The fate of oil clusters during fractional flow: trajectories in
the saturation-capillary number space”, SCA2015-007.
9. Paganin, D., S. Mayo, T. Gureyev, P. Miller and S. Wilkins, "Simultaneous phase and
amplitude extraction from a single defocused image of a homogeneous object."
Journal of microscopy (2002) 206, 1, 33-40.
10. Marone, F., and M. Stampanoni, "Regridding reconstruction algorithm for real-time
tomographic imaging." Journal of synchrotron radiation (2012), 19, 1029-1037.
11. Rezaei, N., and A. Firoozabadi, “Macro- and Microscale Waterflooding Performances
of Crudes which form w/o Emulsions upon Mixing with Brines”, Energy & Fuels
(2014), 28, 2092-2103.
12. Sohrabi, M., P. Mahzari, S. Farzaneh, J. Mills, P. Tsolis, S. Ireland, “Novel Insights
into Mechanisms of Oil Recovery by Low Salinity Water Injection”, This paper was
prepared for presentation at the SPE Middle East Oil & Gas Show and Conference
held in Manama, Bahrain, 8-11 March 2015, SPE-172778-MS.
13. Fredriksen, S., A. Rognmo and M. Fernø, “Pore-Scale Mechanisms During Low
Salinity Waterflooding: Water Diffusion and Osmosis for Oil Mobilization”, paper
prepared for presentation at the SPE Bergen One Day Seminar held in Bergen,
Norway, 20 April 2016, SPE-180060-MS.
14. Avraam, D. and A. Payatakes, “Flow regimes and relative permeabilities during
steady-state two-phase flow in porous media”, Journal of Fluid Mechanics (1995),
293, 207-236.
15. Berg, S., R. Armstrong, A. Georgiadis, H. Ott, A. Schwing, R. Neiteler, N. Brussee,
A. Makurat, M. Rücker, L. Leu, M. Wolf, F. Kahn, F. Enzmann and M. Kersten,
“Onset of oil mobilization and non-wetting phase cluster size distribution”, SCA2014-
022, (2014).
16. Youssef, S., E. Rosenberg, H. Deschamps, R. Oughanem, E. Maire, and R. Mokso,
“Oil ganglia dynamics in natural porous media during surfactant flooding captured by
ultra-fast x-ray microtomograpy”, SCA2014-023, (2014).
17. Georgiadis, S. Berg, A. Makurat, G. Maitland and H. Ott, “Pore-Scale micro-CT
Imaging: Non-Wetting Phase Cluster Size Distribution During Drainage and
Imbibition”, Physical Review E (2013), 88, 3.
 SCA2016-008 1/12

EXPERIMENTAL CARACTERISATION OF THE

RELATIVE GAS PERMEABILITY OF VERY LOW
PERMEABILITY ROCKS
F. Bignonnet1, P. Egermann2, Z. Duan3, L.Jeannin4, F. Skoczylas3
1 : Université de Nantes, GeM équipe IEG, UMR CNRS 6183, France
2 : Storengy (ENGIE group), France
3 : Ecole Centrale de Lille, France
4 : ENGIE E&P International, France

This paper was prepared for presentation at the International Symposium of the Society of Core
Analysts held in Snowmass, Colorado, USA, 21-26 August 2016

ABSTRACT
The relative gas permeability (krg) of a low permeability caprock has been measured on a
set of samples. The intrinsic gas permeability of the samples are typically in the order of
10-4-10-3 mD in the dry state at in-situ confining stress (9 MPa). Water saturations of the
samples are achieved by equilibrating them in chambers whose relative humidities are
imposed. The measured krg points lie on a main curve for all the samples and exhibit a
very high critical gas saturation (Sgc) around 30% above which gas flow can be detected.
The origin of this very specific behaviour is discussed using the percolation theory and an
analogy with the krg curves shape observed during depletion tests. A possible explanation
involves flow of gas as a discontinuous phase at gas saturations below Sgc during the
percolation regime, leading to extremely low krg values with a plateau shape like in
depletion tests.
INTRODUCTION
The characterization of low permeability rocks comprising clay is important in the
context of unconventional reservoir production, basin modelling and geological storage
(CO2, natural gas,). Nevertheless the available literature data about their associated
relative permeabilities are scarce mainly because of the complexity to derive these data in
a representative manner and in a reasonable time frame from laboratory experiments
(Lefort et al., 2011). This work is a contribution to provide such a dataset in a tight
carbonate with a significant clay fraction.
PROCEDURES
Material
The studied material is a caprock, featuring very fine grained carbonate aggregates and
minor clay content. All the samples used in this study have been extracted from a core
taken from a caprock layer located at a depth around 450 m. A confining pressure of 9
MPa has been assumed representative of in-situ conditions for this layer.
 SCA2016-008 2/12

Geochemical analysis indicates that the carbonate content is high with 276,000 parts per
million in weight (ppm) of Ca. Additionally, the rock comprises 7,030 ppm of Mg, 7,980
ppm of K, 11,250 ppm of Fe and 6,130 ppm of S, which indicates silicates and clay
content. The total organic carbon is 0.57 % and the pyrite component is minor (1.1 %).
The anions have not been measured in the “bulk” geochemical analysis, but local EDX
analysis indicates that the carbonate is made of tightly packed aggregates of size 1-10
µm, separated by thin layers or canaliculi. The clay content - most likely illite - is located
in these thin layers surrounding the carbonate aggregates as well as in isolated high-
porosity pockets of micro-metric extension which exhibit micro-metric porosities.
Methods
Sample preparation: Eight vertical cylindrical samples with diameter 37 mm and height
10 mm have been extracted from the main vertical core. The axis of the cylinders is
assumed perpendicular to the bedding. The cylindrical samples are first obtained by 37-
mm coring, and then cut to the specified length. The faces of the samples are then
polished with sandpaper. The height h and diameter d of each sample are then measured
with a calliper with a precision of 0.02 mm. The total volume V of each sample is simply
assessed by V = πhd2/4.
Water Sorption Isotherms and Water Porosity Measurements
Step 1: Drying. All samples are first oven-dried at 60oC until mass stabilisation is
achieved, which occurs after about two weeks. This temperature has been chosen to
ensure a fairly good state of drying, while limiting the damage (e.g. micro-cracking),
which could arise at higher temperatures. This state is considered as the reference dry
state, and the water saturation Sw is assumed null. The mass in the dry state m dry is
measured with a precision of 0.01g.
Step 2: Water saturation. To obtain a partial water saturation, the samples are stored in a
chamber comprising air and water vapour in equilibrium with a saturated saline solution
at 20oC (Chen et al., 2012). The relative humidity of the atmosphere in the chamber is
imposed by thermo-dynamical equilibrium and depends on the type of saline solution.
The relative humidities imposed in the present study range from 43% to 98%.In turn, the
capillary pressure Pcap = Pg − Pl between the gas (here at atmospheric pressure) and the
liquid water (which condenses in the pore network of the sample) is related to the relative
humidity RH of the atmosphere by Kelvin’s law (first part of equation 1). The capillary
pressure is itself related to the curvature of the liquid-gas menisci by Young-Laplace’s
law (second part of equation 1). If the contact angle between the water and the solid
phase is null, the two relations imply:
ρl RT ln(1/RH) / M = Pcap = 2γ / r (1)
where ρl is the density of water, M the molar mass of the water, R the perfect gas
constant, T the temperature, γ is the coefficient of surface tension, (γ=0.073 N/m for air
and water under ambient conditions) and r is a pore size. More precisely, r is either the
radius of a cylindrical pore or the distance between the two walls of a flat pore. The
capillary pressure thus increases as the relative humidity decreases. For water at 20oC, the
 SCA2016-008 3/12

capillary pressures corresponding to the imposed relative humidities range from 2.7 MPa
for RH = 98% (corresponding to r=53nm) to 114 MPa for RH = 43% (corresponding to
r=1.3nm).
A sample successively equilibrated in chambers with increasing relative humidity
undergoes a progressive imbibition, from the smaller to the larger pores. At each step,
equilibrium is assumed once mass stabilisation of the sample is achieved, which occurs
after about two weeks. The mass m(RH) in the partially water saturated state due to the
relative humidity RH is measured with a precision of 0.01 g, and will be used after the
third step to deduce the partial water saturation.
Step 3: Total Water Saturation. At the end of the previous steps, the samples are water
saturated. To do so, the samples are placed in a vacuumed chamber for 24 hour ; they are
partially immersed in water for the 6 first hours to allow the air to drain, then fully
immersed. The mass in this saturated state msat is measured. Assuming the final
immersion step guarantees a full water saturation, the porosity φ is:
φ = (msat − mdry) / (ρl V) (2)
Porosity measured using this method will be referred to as “water porosity”, as opposed
to the “gas porosity” to which it will be compared to later in the paper. Additionally, the
partial water saturation Sw(RH) achieved at any of the previous steps by equilibrium with
relative humidity RH may now be assessed using
Sw(RH) = (m(RH) − mdry) / (msat − mdry) (3)
The Sw − RH curve obtained from equation 3 can be used to derive the capillary pressure
curve by means of Kelvin’s law (equation 1, left). The latter may be used with caution to
estimate the pore size distribution through Young-Laplace’s law (equation 1, right),
provided the pores are assumed to be either cylindrical or flat and connected in a
hierarchy from the smallest to the largest.
Gas Porosity Measurements
An alternative porosity measurement method has been also used using a dedicated
method (Chen et al., 2013; Nadah et al., 2013) similar in principle to gas pycnometry.
The measurement is performed on a dried sample in a confining cell and relies on the
injection of 99% pure Argon in the pore network. The room temperature is set to 20oC.
Three volumes have to be distinguished: (1) a reference tank with a calibrated volume Vt
of 70 mL, (2) the pore volume φ × V to be measured, (3) the volume Vc of the tubes
connecting the reference tank to both sides of the sample. The reference tank is initially at
a pressure Pt of 1 MPa which is monitored using a pressure sensor with a precision of 100
Pa, whereas the porosity and the connecting tubes are at atmospheric pressure Patm . The
valves between the reference tank and the sample are then opened while the pressure in
the reference tank is monitored. Once gas pressure and temperature are equilibrated,
application of the law of perfect gases in isothermal conditions links the final pressure P f
to the sample porosity through:
P t Vt + Patm (Vc + φV) = Pf (Vt + Vc + φV) (4)
 SCA2016-008 4/12

The porosity measurement deduced from equation 4 is referred to as “gas porosity”. The
volume Vc (≈ 5 mL) is measured based on the same principle using a non-porous steel
sample.
Gas Permeability Measurements
Gas permeability K is measured using a uniaxial steady state gas flow apparatus. The
apparatus consists of a confining cell which is allowed to reach confining pressures Pc as
high as 60 MPa, together with a gas injection device (Dana and Skoczylas, 1999, 2002;
Lion et al., 2004, Davy et al., 2007; Chen et al., 2012, 2013; Nadah et al., 2013; Duan et
al., 2014). The gas used in the current study is 99% pure Argon with a viscosity of µ =
2.2×10−5 Pa.s.
The injection pressure Pi at the upstream side is set to values ranging from 0.5 MPa to 1.5
MPa, whereas the back-pressure Pd at the downstream side is equal to the atmospheric
pressure. The injection pressure is measured with a pressure sensor with a precision of
0.002 MPa. The drainage flow rate qd at x = h is measured with a set of flow-meters, the
finer of them allowing measurement of flow rates from 0.1 ml/min to 1 ml/min with a
precision of 0.01 ml/min.
Combining Darcy’s law, mass conservation and the law of perfect gases to Klinkenberg's
law (1941) to account for slip flow effects yields the expression for the apparent gas
permeability Kapp as a function of the measured drainage flow rate qd in the steady state
regime:
Kapp = Kint(1+b/Pm) = 2µhPdqd / A(P i2−P d2) with Pm = (Pi +P d)/2 (5)
where Kint is the intrinsic gas permeability, b is Klinkenberg’s coefficient and A = πd2/4
is the area of the sample section. The equation 5 has been rigorously derived accounting
for the non-uniformity of the pressure gradient profile along the sample due to the gas
compressibility and is valid irrespective of Pi and Pd, which may be very different. Hence,
the plot of Kapp with respect to 1/P m allows determination of the intrinsic gas
permeability and the Klinkenberg coefficient. In practice, three different gas injection
pressures are successively applied: 0.5, 1 and 1.5 Mpa.
In order to assess the relative gas permeability as a function of the water saturation Sw,
effective gas permeability measurements are performed at each step of saturation
obtained by the procedure described above. As the Argon gas injected in the partially
water saturated sample is dry, the mass of the sample is checked before and after the
measurement to monitor potential desaturation of the sample. In practice for the current
study, changes in water saturation due to dry gas injection have been observed to be
negligible.
RESULTS
Porosity
The measured “water porosities” (equation 2) and “gas porosities” (equation 4) are
reported in Table 1. Water porosities are measured without confining pressure, whereas
 SCA2016-008 5/12

gas porosities are measured at three steps of a confining pressure cycle (from 2 MPa to 9
MPa (≈ in-situ pressure) and back to 2 MPa).
confining pressure (MPa)

0 2 → 9 → 2 2 9 2 9 2 9

porosity permeability
sample water gas Kapp (µD) Kint b
(Argon) for Pi = 1MPa (µD) (MPa)

MIL-12 10.6 - - - 1.44 1.28 0.28 0.23 2.4 2.8

MIL-11 9.6 11.84 10.67 11.53 1.20 1.12 0.18 0.16 3.4 3.6

MIL-14 8.9 12.43 11.69 12.15 1.21 1.13 0.19 0.18 3.3 3.2

MIL-20 14.5 - - - 9.41 3.94 7.0 2.4 0.23 0.40

MIL-21 13.3 - - - 2.15 1.83 0.55 0.44 1.7 1.9

MIL-22 10.2 12.70 11.79 12.50 1.36 1.26 0.42 0.38 1.1 1.0

MIL-23 9.0 11.65 10.81 11.51 1.13 1.05 0.18 0.16 3.3 3.3

MIL-24 9.1 11.75 11.13 11.77 1.16 1.09 0.19 0.21 3.0 2.5
Table 1. Measured porosities and dry-state gas permeabilities.
Gas porosities exhibit a low dispersion and range from 11.7 to 12.7% at 2 MPa of
confining pressure. Water porosities are more dispersed, mostly because of the presence
of the two specimens MIL-20 and MIL-21, which have been sampled closer to the upper
part of the received core than all other samples.
The water porosities appear to systematically underestimate the gas porosities by 1 to 2
porosity units. As water molecule and Argon atom diameters are respectively 0.28 nm
and 0.38 nm, their size cannot explain this discrepancy. The possible origin of this
difference may be the sorption of Argon on pore walls during the porosity measurement
experiment, which has not been quantitatively assessed.
The confining pressure increase from 2 MPa to 9 MPa leads to a decrease of 1 porosity
unit. After unloading, the initial porosity values are retrieved up to the measurement
errors. This indicates that the evolution of the samples along the imposed stress path is
mostly elastic.
Water Sorption Isotherm, Capillary Curve and Pore Size Distribution
In order to determine the length scales associated with the porosity, the water sorption
and desorption curves are measured based on the partial water saturation method
described in the previous section. Note that the Sw−RH curves obtained from equation 3
 SCA2016-008 6/12

Figure 1. Water sorption-desorption isotherms (left) and inferred cumulative pore size
distribution from Kelvin-Laplace's law (right).
and presented in Figure 1 can be used to derive capillary pressure curves by means of
Kelvin’s law (equation 1).
The desorption curve, which has been measured on a single sample, is slightly above the
sorption curves. As for the set of eight sorption curves, the dispersion increases with
decreasing capillary pressure (Sw = 0.68 ± 0.10 at Pcap = 2.7 MPa ; Sw = 0.115 ± 0.015 at
Pcap = 114 MPa). Again, the two specimens MIL-20 and MIL-21, which have the highest
total porosities, have also the weakest water saturation at Pcap = 2.7 MPa. This indicates
that the extra porosity of these two specimens is mostly a macro-porosity.
The sorption-desorption curves presented in Figure 1 are re-interpreted as cumulative
pore size distributions using Young-Laplace’s law (equation 1). To do so, the pores have
been assumed flat, i.e. described as the space between two parallel planes with an
aperture r, which is of course a rough approximation. Under these assumptions, around
70±5% of the porosity corresponds to pores with an aperture below 50 nm. This fine
porosity is attributed to the clay phase, which is located either in the canaliculi spanning
carbonate aggregates or in high-porosity pockets. The pore size distribution above 50 nm
cannot be evaluated by the present method, but mainly corresponds to the inter-particular
pores observed by SEM in the high-porosity pockets.
Gas Permeability
Dry case
The gas permeability has been initially measured on the dried samples using the steady
state method. Each sample undergoes a confining pressure cycle from 2 MPa to 9 MPa
and back to 2 MPa. At each confining pressure step, the apparent gas permeability is
measured for three different gas injection pressures: 0.5 MPa, 1 MPa and 1.5 MPa.
Except for the two samples MIL-20 and MIL-21, the apparent gas permeability for an
injection pressure of 1 MPa is 1.25±0.15 µD and Klinkenberg corrected intrinsic
permeability is 0.24±0.1 µD (see Table 1). The specimens MIL-20 and MIL-21, which
 SCA2016-008 7/12

also have more micro-cracks and macro-porosity, have permeabilities 2 to 8 times greater
than all other samples.
The moderate increase in confining pressure from 2 to 9 MPa results in a decrease in
permeability of only 10%. After unloading to 2 MPa, the initial permeability is almost
recovered with less than 5% relative difference. These relatively low fluctuations indicate
that the deformation of the samples is mostly elastic during the present load cycle.
Relative gas permeability krg
In order to assess the relative gas permeability, effective gas permeability measurements
are carried out on the partially water saturated samples. At each imposed relative
humidity step, the samples undergo the same confining pressure/gas injection pressure
cycles as in the dry case.
The measured gas relative permeabilities are reported as a function of water saturation in
Figure 2, for confining pressures of 2 and 9 MPa.

Figure 2. Gas relative permeability krg.

For water saturations below 0.5, effective gas permeabilities exhibited limited hysteresis
to a cycle in confining pressure (2→9→2 MPa). However at the highest water saturation
tested (corresponding to RH = 98%), the effective gas permeability after the confining
pressure cycle ranged only from 1/2 to 1/3 of the value before the cycle, for the same
reference value (2 MPa) of confining pressure. Two hypotheses may be proposed to
explain this hysteresis: (1) pore compression due to hydrostatic loading induces a spatial
reorganisation of the water or (2) the saturation cycle has induced micro-cracking of the
samples. Subsequent permeability measurements after a second drying of the samples
suggest that limited micro-cracking might indeed have occurred.
Except for the two outlier specimens MIL-20 and MIL-21 which suffered from severe
micro-cracking induced by the saturation process, all the measured relative permeabilities
in Figure 2 lie on a master curve, for both confining pressures (2 and 9 MPa). The main
feature of the relative permeability curve is the presence of a high critical gas saturation
Sgc around 0.29 ± 0.04 for which gas permeability nearly vanishes. The influence of the
 SCA2016-008 8/12

saturation history on the relative permeability curve has not been assessed and requires
further investigation. This effect has been studied on two industrial concretes (Chen et
al., 2012), evidencing that the saturation history influences krg mainly in the case of a
strong hysteresis in desorption/re-saturation isotherms (i.e. Sw − RH relations), which is
not the case of the material presently under consideration (Figure 1).
DISCUSSION
Are the krg curves obtained relevant for drainage or imbibition ? A priori the saturation
protocol relies on a progressive imbibition through the relative humidity technique.
Nevertheless, the trapped gas phase is compressed from 1 atm (initial pressure) to very
large pressures (Young Laplace) since the pressure is imposed by the size of the throat
which trapped the gas. For 50 nm, it gives a pressure of around 30 bars leading to a
strong compression and a dissolution of the gas phase in the surrounding water phase. It
makes this protocol be an equivalent drainage process since all the gas trapping is
progressively and naturally removed by this compression-dissolution mechanism.

Figure 3. Examples of gas drainage relative permeability curves obtained on tight reservoirs, Cluff and Byrnes,
(2010) on the left and Wang et al (2011) on the right)

The gas drainage relative permeability obtained on a caprock facies of extremely low
permeability, appear very similar to the ones already published for tight gas reservoirs
since they exhibit also very high Sgc values (typically in the 15-40% range) as shown in
Figure 3 (Cluff and Byrnes,2010; Wang et al., 2011 and Duan et al., 2014). The Sgc is
defined here as the saturation which corresponds to the apparent beginning of a gas flow
in a connected form. Cluff and Byrnes (2010) proposed an empirical formula based on
tight reservoir sample to obtain the apparent Sgc value (Sgc = 0.15 – 0.05*log10K). This
does not seem to be applicable for our lower permeability samples since it would give
more than 40% rather the observed values in the 25-30% range.
Above the Sgc value, the flow topology (connected gas phase) is consistent with the
Darcy formalism and therefore a Corey type function appears adequate to reproduce the
 SCA2016-008 9/12

general shape of the data (Figure 5). Surprisingly, the gas relative permeability behaviour
below Sgc has raised little interest to date (Lefort et al., 2011). It is often considered that
the gas phase exhibits no mobility at all in this saturation range, which is often referred as
the “permeability jail”. As pointed out by Cluff and Byrnes (2010), this is not true in a
drainage mode by flooding since the gas must have some mobility to make its own
saturation increase up to Sgc. The expected shape of the krg curve in this saturation
interval have been inferred by combining several sources of information:
1. Although this parameter is difficult to measure directly, both Busch and Amann-
Hildenbrand (2013) and Haoguang et al. (2014) have reported that very low
permeability rocks are characterized by large aspect ratios (pore body size divided by
pore throat size). On the opposite, the length ratio (pore body length divided throat
length) is small in this particular rocks (elongated throats).
2. Several interesting results can be obtained from the works on the percolation-invasion
mechanisms (Lenormand et al., 1983; Wilkinson, 1986). In the percolation regime at
low saturation, the krg is proportional to Sg . α is in the order of 5 and depends on the
α

conductivity exponent of the medium. It is also established that the aspect ratio plays
a major role in the snap-off frequency during the percolation. It therefore suggests
that the texture of low permeability rocks significantly favours this mechanism. This
could induce the formation of disconnected gas ganglia and a dispersed flow topology
at a certain level of gas saturation. This mechanism could be potentially exacerbated
by the small length ratio which also prevails in very low permeability rocks. In the
configuration of a drainage at low rate with a low viscosity gas phase, the DLA
(Diffusion Limited Aggregation) percolation mode is the one which prevails
(Lenormand et al., 1988). These snap-off events could occur not only at the flood
front but all along the DLA flow path leading to a dispersed, disconnected gas
distribution pattern and a reduced gas mobility comparing to what predicts the
percolation theory.

Figure 4. Examples of solution gas relative permeability curves obtained with light oils (Egermann et al., 2004) and heavy
oils (Tang et al., 2006)
 SCA2016-008 10/12

3. The dispersed flow characteristics have been extensively studied in the context of the
recovery processes based on solution gas drive by depletion under secondary / tertiary
conditions for both light / heavy oils (Egermann et al., 2004, Bauget et al., 2005,
Tang et al., 2006). In these cases, the discontinuous gas phase is created in-situ by
nucleation and then diffusion from the solution gas. The DLA – snap-off events
suspected in low permeability rocks could conduct to a similar gas distribution pattern
once the gas saturation reached a certain value. All the results from the studies on
solution gas drive mechanisms converge towards a very typical shape of the krg
curves with a plateau at very low values (Figure 4). Whatever the type of fluid system
considered, the plateau is typically located below 10-3-10-4.
By combining these various sources of information, it is then possible to propose what
the global krg shape could be over the whole saturation range. At low saturation (Figure 5
left red curves), the percolation krg curve is quite steep rather the solution gas drive like
krg curve exhibits a plateau ending at around 10-4. This value is completely speculative
and has been chosen only because it seems representative of the values found for similar
viscosity ratio cases. Additional studies would be needed to better assess this value
according to the rock and fluid/system. At very low Sg, the percolation krg curve is more
relevant (see paragraph 3 above). At higher Sg, a dispersed gas flow regime could start
before Sgc and then the connected flow (blue part). The proposed shape of krg below the
apparent Sgc is a composite curve built from percolation and dispersed flow theory
(Figure 5 right green curve). It is interesting to notice that this composite curve is
normally completely hidden because the measurement of such low values of krg are not
accessible with standard protocols with low permeability rocks.

Figure 5. Possible shape of the global krg curve over the whole saturation range for very low K rocks

Having the complete shape of the krg is of interest for various industrial applications
especially those with large time scales, which give more importance about the role played
by low krg values. It concerns mainly the applications related to the basin modelling
simulation; the CO2 storage or the nuclear waste storage (evolution of the associated H2
produced) where accurate and complete krg curve description is needed to assess the long
fate of the gas phase.
 SCA2016-008 11/12

The specific krg curve shape has also strong implications for gas storage applications
since it implies that the gas mobility is very limited at low saturation. It means that a
caprock exhibits two interesting features in term of confinement properties. The first is
the standard one which consists in a static barrier related to the threshold capillary
pressure value. The second one, evidenced by this work, consists in a dynamic barrier,
related to the specific gas relative permeability curve shape, which delays significantly
the gas flow at low saturations.
CONCLUSION
This paper provides a new complete set of experimental data at various confining
pressures to characterize a caprock facies including porosity accessible to gas or water,
water sorption-desorption isotherms and gas permeabilities as a function of water
saturation, gas injection pressure and confining pressure. The measured relative gas
permeabilities lie on a main curve for all samples and exhibit a very high critical gas
saturation around 30% where gas flow can be detected. This behaviour is very specific
and was already evidenced on tight reservoir samples. Several pieces of information
about the pore texture in low permeability rocks, the percolation-invasion mechanisms
and the works conducted for determining flow parameters under solution gas drive, have
been combined to assess the flow behaviour below this apparent Sgc value. A composite
curve is proposed which exhibits extremely low gas relative permeability far below the
limit of what can be measured in the laboratory in such rocks. When the gas saturation
becomes large enough, a connected flow can take place and be measured. This specific
behaviour consists in a key feature for various industrial applications with large
characteristic times since it rules the flow kinetic in this low gas saturation range and krg
values.
ACKNOWLEDGEMENTS
We are grateful to ENGIE E&P International and Storengy (ENGIE group) for
permission to publish these results and to R. Lenormand for fruitful discussions.
REFERENCES
F. Bauget, P. Egermann, R. Lenormand, “A new model to obtain representative field
relative permeability for reservoirs produced under solution gas drive”, SPEREE, (August
2005), 348-356.
A. Busch, A. Amann-Hildenbrand, “Predicting capillarity of mudrocks”, Marine and
Petroleum Geology, (2013) 45, 208-223.
W. Chen, J. Liu, F. Brue, F. Skoczylas, C.A. Davy, X. Bourbon, and J. Talandier, “Water
retention and gas relative permeability of two industrial concretes,” Cement and Concrete
Research, (2012) 42, 1001–1013.
X.T. Chen, G. Caratini, C.A. Davy, D. Troadec, and F. Skoczylas, “Coupled transport
and poro-mechanical properties of a heat-treated mortar under confinement”, Cement and
Concrete Research, (2013) 49, 10–20.
 SCA2016-008 12/12

R.M. Cluff, A.P. Byrnes, “Relative permeability in tight gas sandstone reservoirs – The
“permeability jail” model”, SPWLA 51th Annual Logging Symposium, (June 19-23,
2010).
E. Dana and F. Skoczylas, “Gas relative permeability and pore structure of sandstones”,
International Journal of Rock Mechanics and Mining Sciences, (1999) 36, 613–625.
E. Dana and F. Skoczylas, “Experimental study of two-phase flow in three sandstones. I.
Measuring relative permeabilities during two-phase steady-state experiments”,
International Journal of Multiphase Flow, (2002) 28, 1719–1736.
C.A. Davy, F. Skoczylas, J.D. Barnichon, and P. Lebon, “Permeability of macro-cracked
argillite under confinement: Gas and water testing”, Physics and Chemistry of the Earth,
(2007) 32, 667–680.
Z. Duan, C.A. Davy, F. Agostini, L. Jeannin, D. Troadec, and F. Skoczylas, “Gas
recovery potential of sandstones from tight gas reservoirs”, International Journal of Rock
Mechanics & Mining Sciences, (2014) 65, 75–85.
P. Egermann, S. Banini, O. Vizika, “Depressurization under tertiary conditions in the
near-wellbore region: experiments, visualization and radial flow simulations”,
Petrophysics, (Sept-October 2004) 45, n°5.
W. Haoguang, M. Kun, Y. Xiang’an, W. Xinxin, “The relationship of ultra-low
permeability sandstone aspect ratio with porosity, permeability”, Advances in Petroleum
Exploration and Development, (2014) 7, n°1, 7-12.
L. J. Klinkenberg, “The permeability of porous media to liquids and gases”, In API
Drilling and Production Practices, New York, API 11th mid-year meeting, Tulsa (1941),
200–213.
R. Lenormand, C. Zarcone, A. Sarr, “Mechanisms of the displacement of one fluid by
another in a network of capillary ducts”, J. Fluid Mech., (1983) 135, 337-353.
R. Lenormand, E. Touboul, and C. Zarcone, “Numerical models and experiments on
immiscible displacements in porous media”, J. Fluid Mech., (1988) 189, 165-187.
M. Lion, F. Skoczylas, and B. Ledésert, “Determination of the main hydraulic and poro-
elastic properties of a limestone from Bourgogne, France”, International Journal of Rock
Mechanics & Mining Sciences, (2004) 41, 915–925.
J. Nadah, F. Bignonnet, C. A. Davy, F. Skoczylas, D. Troadec, and S. Bakowski,
“Microstructure and poro-mechanical performance of Haubourdin chalk”, International
Journal of Rock Mechanics & Mining Sciences, (2013) 58, 149–165.
G.-Q. Tang, A. Sahni, F. Gadelle, M. Kumar, A.R. Kovscek, 'Heavy-oil solution drive in
consolidated and unconsolidated rock”, SPE Journal, (June 2006), 259-268.
D. Wilkinson, “Percolation effects in immiscible displacement”, Physical review A,
(1986) 34, n°2, 1380-1391.
P. Lefort, V. Sarrot, M. Prat, J. Talandier, “Hyperslow drainage”, 20ème congrès français
de mécanique, 29 août-2 septembre 2011, Besançon, France.
Y. Wang, Z. Chen, V. Morah, R. J. Knabe and M. Appel. Gas phase relative permeability
characterization on tight gas samples. SCA2011-13.
 SCA2016-016 1/12

MEASUREMENTS OF ROCK CORE DISPERSIVITY AND

TORTUOSITY FOR MULTI-PHASE SYSTEMS
Marco Zeccaa, Abdolvahab Honaria, Sarah J. Vogta, Branko Bijeljicb, Eric F. Maya and
Michael L. Johnsa
a
School of Mechanical and Chemical Engineering M050, University of Western Australia,
35 Stirling Highway, Crawley 6009, Western Australia, Australia.
b
Department of Earth Science and Engineering, Imperial College London,
Prince Consort Road, London SW7 2BP, United Kingdom.

This paper was prepared for presentation at the International Symposium of the Society of Core
Analysts held in Snowmass, Colorado, USA, 21-26 August 2016

ABSTRACT
Known as CSEGR (Carbon Sequestration with Enhanced Gas Recovery), sequestered CO2
can be used to re-pressurise partially depleted natural gas reservoirs to increase gas
recovery, with the intention to also store the CO2. Because natural gas and CO2 are fully
miscible, simulations require CO2-natural gas dispersion at supercritical conditions to be
accurately measured in order to quantify this mixing and hence establish the viability of
CSEGR. We have designed and constructed a unique core flooding apparatus capable of
accurately determining such supercritical gas dispersivity; this has been measured in two
sandstones and two carbonates cores as a function of temperature, pressure and interstitial
velocity. This is supplemented with Nuclear Magnetic Resonance (NMR) measurements of
tortuosity to quantify the underpinning effective diffusion process. The resultant
dispersivity data (K - Peclet number (Pe) plot) and tortuosity data are combined and
presented in a format which is readily incorporated into CSEGR simulations.

INTRODUCTION
Even though enhanced oil recovery (EOR) via CO2 flooding is a promising method to
improve the recovery factor of oil reservoirs and geologically sequester CO2, enhanced gas
recovery (EGR) with CO2 flooding/sequestration has not been widely considered by the oil
and gas industry. The only current field-scale EGR project is the Rotliegend K12-B gas
reservoir, located offshore of the Netherlands, which started in 2004 after 17 years of
conventional gas production (Vandeweijer et al., 2011). A Canadian depleted gas reservoir
was also used for an EGR/CO2 sequestration trial in 2002 but the operation was terminated
after three years due to the early CO2 breakthrough into the producing wells (Pooladi-
Darvish et al., 2008).
 SCA2016-016 2/12

EGR cannot only safely store CO2 within the formation but also can improve the natural
gas recovery by maintaining the reservoir pressure and enhancing sweep efficiency and
production rates. However, natural gas and CO2 are entirely miscible in all proportions and
consequently the risks of extensive mixing of these fluids within the reservoir formation
and early CO2 breakthrough into production wells are the main hurdles for EGR
implementation. These associated risks and uncertainties of EGR projects can be estimated
by using reservoir simulations in which fluid flow in reservoir formations can be
quantitatively modeled and, in principle, the mixing process between the injected CO2 and
natural gas captured. Doing so reliably, however, requires (1) adequate characterization of
physical dispersion at all relevant length-scales, and (2) that numerical dispersion in such
simulations be kept sufficiently small. Extensive studies of dispersion at the field scale
have been published (Arya et al., 1988; Coats et al., 2009a; Lake, 1989) but generalizing
the results of these studies is not straightforward. Physical dispersion at smaller length
scales is also important because the description of transport at the core scale can play an
important role in the prediction of mixing at the field scale.

In this work, a unique core flooding apparatus capable of accurately determining

supercritical gas dispersivity has been designed and constructed. The measured dispersion
coefficients (KL) of CO2 and CH4 for two sandstone cores (Berea and Donnybrook) and
two carbonate rock samples (Ketton and Estaillades) are presented over a range of
pressure, temperature, and interstitial velocities. These highly reproducible data are used to
obtain the characteristic mixing length-scale for the different rocks. These measurements
were extended to also consider the effect of residual water on the dispersion process and
were supplemented by an independent measurement of tortuosity using appropriate NMR
methodology.

THEORY
Dispersion in floods of cylindrical core plugs with injection along the axis can be
adequately described by the one-dimensional advective-dispersion (AD) equation:
∂C ∂ 2C ∂C
= K L 2 − um , (1)
∂t ∂x ∂x
where C is the concentration of the dispersing species, t is time, KL is the longitudinal
dispersion coefficient and um is the mean interstitial velocity in the direction of bulk flow,
x, in the porous medium. We note the use of alternative models (continuous time random
walks and multi-rate mass transfer models) to describe dispersion in more complex,
heterogeneous porous media (e.g. Haggerty and Gorelick, 1995; Berkowitz et al., 2006).
The ratio of the dispersion coefficient calculated from Eq. (1) to the diffusion coefficient,
KL/D, has been commonly plotted as a function of Peclet number. This is defined here as
αum/D where α denotes the medium dispersivity. This differs from the classical definition
in which grain/particle diameter replaces α as the characteristic length scale for mixing in
the pores (Perkins and Johnston, 1963). Note that this latter definition is inappropriate for
consolidated media such as the rock cores studied here.
 SCA2016-016 3/12

Additionally, Eq.(2) has been frequently used to characterize the different dispersion
behaviors evident in such correlated data:
K L 1 α umn
= + , (2)
D τ D
where τ is tortuosity, and n is an exponent. These three parameters characterize a given
porous medium: generally 1 ≤ n < 1.5 (Brigham et al., 1961; Gist et al., 1990; Legatski and
Katz, 1967) and τ can range from 2 (for ideal packed beds only) to as much as 13 for
certain consolidated media (Gist et al., 1990; Hurlimann et al., 1994). Generally for
consolidated media these three parameters are treated empirically and must be determined
from experiments with the rock; while τ can be determined via several methods, α and n
can only be determined from core flooding experiments.

METHODOLOGY
Materials
The methane, carbon dioxide and nitrogen used in our experiments were supplied by BOC
Scientific at purities greater than 0.999 mole fraction. Mixtures of CO2 and CH4 for pulse
injection were prepared gravimetrically. Two types of 1.5 inch diameter sandstone cores,
Berea and Donnybrook, and two carbonates, Estillades and Ketton were available. These
cores were cleaned prior to use by Soxhlet extraction with a mole fraction mixture of 0.9
dicholormethane (Chem-Supply) + 0.1 methanol (Ajax-Finechem). The geometrical
properties of the four cores used are listed in Table 1 with porosities, Φ, and
permeabilities, k.

Length Diameter pconfining

Core Φ k (mD)
(mm) (mm) (MPa)
8 0.2043 463.3
Berea sandstone 100.4 37.6 10 0.2039 460.7
12 0.2036 458.2
8 0.1576 12.07
Donnybrook
101 37.5 10 0.1573 12.01
sandstone
12 0.1572 11.95
8 0.2277 2922.4
Ketton carbonate 104.7 38.0 10 0.2274 2912.2
12 0.2270 2902.7
8 0.2817 211.7
Estillades carbonate 105.6 37.9 10 0.0811 211
12 0.2806 210.1
Table 1 Rock cores properties.
 SCA2016-016 4/12

Apparatus and method

The specialized core flooding apparatus for SCCO2-CH4 dispersion measurements was
adapted from our previous experimental set-up for breakthrough measurements (Hughes et
al., 2012) to allow for pulse injection and to enable water delivery to, and drainage of, the
rock core. A schematic of the apparatus is shown in Figure 1. A HPLC switching valve
was included to allow for a controlled injection of a pulse of a 50% CH4 and 50% CO2
mixture. This composition was calculated as the best trade-off between minimizing
density and viscosity contrasts between the fluids and maintaining an adequate signal to
noise ratio. Analysis of the core effluent was conducted at measurement pressure using a
HPIR cell and a syringe pump was used for effluent collection, ensuring better pressure
regulation during the pulse displacement process. Details of the core holder, core
wrapping, the core holder bath temperature control, pressure instrumentation and
calibration of FT-IR spectrometer are included in our previous work (Hughes et al., 2012).
All experiments described below were conducted with the core in a vertical orientation.

ΔP N2

P1 P2
T V5
V3
V4

Switching
V1 Valve
Infrared
Triaxial Core Holder

Spectrometer
Core
0

Injection V2 Collection
Syringe Pump Syringe Pump

V8
CH4 Vent Silicone Oil Silicone Oil
Bath
Metering Pump
Pump
Separator

CH4+CO2
mixture V6
V9

V7 Drain
Scale
Vacuum pump Brine/Condensate Pump

Figure 1 Simplified schematic of the core-flood apparatus for accurate dispersion measurements. The HPLC
switching valve is shown in the sample loop filling position. To inject the pulse the valve is switched (rotated
180◦) so that the CH4 flow flushes the sample out of the loop.

The flow from the injection pump was stopped when the dispersed pulse had passed
completely through the IR cell. The experiments were conducted at core temperatures and
pressures of 40 to 100 °C and 8 to 14 MPa, respectively, and with interstitial velocities
between 0.004 and 1.908 mm·s-1.
 SCA2016-016 5/12

Data analysis
For pulse experiments, the boundary conditions are:
C(x > 0, t = 0) = 0, C(x = ∞, t) = 0, C(x = 0, 0 < t < δ) = C0 and C(x = 0, t > δ) = 0 where δ
is the pulse duration. The solution to Eq. (1) with these boundary conditions (Levenspiel,
1999) is:
um ⎛ ( L − um t )2 ⎞
C= exp ⎜ − ⎟, (3)
4π K Lt ⎜
⎝ 4 K L t ⎟
⎠

where L is the experimental length scale (core length). Eq.(1), and its solution (Eq.(3)) are
one–dimensional, as such they are unable to account for any velocity variations in the
radial or cross-sectional direction.

However, Eq.(3) alone was not adequate for description of the effluent tracer concentration
curves produced from the Ketton carbonate because of the non-Fickian/preasymptotic
transport behavior associated with its heterogeneity in pore sizes. To describe this non-
Fickian flow regime, several alternative models have been developed including the mobile-
immobile model (MIM) (Deans (1963)), diffusion models, MRMT models (Haggerty and
Gorelick, 1995) and CTRW models (Berkowitz et al., 2006). In this case, it was found that
the non-Fickian behavior of Ketton carbonate could be adequately modeled utilizing the
MIM approach, which assumes the existence of two distinct regions in the rock, mobile
and stagnant/immobile, and which describes the diffusional transport of species between
these two zones using a first order mass transfer expression. This model was initially
proposed by Deans (1963), who added two new parameters to Eq.(1), namely the mass
transfer coefficient and immobile volume fraction, but did not include the longitudinal
dispersion coefficient. Coats and Smith (1964) modified Deans’ version of Eq.(1) by
adding this dispersion coefficient to produce (Coats and Smith, 1964; Van Genuchten and
Wierenga, 1976)

∂Cm ∂Cim ∂ 2Cm ∂C

θm + θim = θm KL 2
− umθm m , (4)
∂t ∂t ∂x ∂x
∂C
θim im = β ( Cm − Cim ) (5)
∂t

where Cm and Cim are the concentrations of the dispersing solute species in the mobile and
immobile regions, respectively; θm and θim are the mobile and stagnant fractions of the
fluid in the porous media; β is a mass transfer coefficient; and um is the mean interstitial
velocity in the mobile zone. To apply this solution to the measured effluent pulse profiles
obtained for the Ketton core, the dispersion coefficient (KL), the mobile fluid fraction (θm)
and the mass transfer coefficient (β) were treated as the fitting parameters and the
measured pulse breakthrough profiles were regressed to the analytical solutions of Eq. (4)
and Eq.(5).
 SCA2016-016 6/12

The dispersion that occurs in the tubing leading to and from the core, and the
inhomogeneous velocity profiles around the core entry and exit, both add to mixing and
erroneously increase the apparent dispersion. As discussed in our previous work (Hughes
et al., 2012), to remove these effects measurements were conducted at the same conditions
of T, P and flow rate with short and long Berea rock cores. The concentration profiles
collected with the short core were used as inlet boundary conditions to a hypothetical
undisturbed core of a length equal to the difference in length between our long and short
core. Eq.(1) was solved numerically using a central finite difference method (the method
of lines) implemented in Matlab within this hypothetical core and regressed to the
experimental (long core) data in order to determine Kcorr. In this manner this systematic
error, between Klong extracted via Eq.(3) and the above methodology employing long and
short core, was quantified. The modified setup produced a correction which was
significantly dependent on velocity; a clear trend is shown in Figure 2. Consequently all
Klong data measured for all the rock cores were corrected to account for these systematic
errors as shown in Figure 2 to give the Kcorr values listed in Table 2.

Figure 2 Relative size of dispersion coefficient correction as function of the interstitial velocity. The two
data sets correspond to the two halves obtained from the cut of the original long core.

Independent measure of tortuosity

The diffusion coefficients D in Eq.(2), which depend on pressure, temperature and
composition, were obtained from the data of Takahashi and Iwasaki (1970). An error in the
calculation of D would cause an offset in the value of (KL/D) determined from the
dispersion measurements, which would be most pronounced in the limit um → 0. To
confirm that no such offset was present, we measured the core’s tortuosity, τ, using a
second, completely independent technique. The Berea and Donnybrook rock core samples
were evacuated and then saturated with distilled water at a pressure of 10 MPa and left for
several hours to ensure complete saturation. These saturated cores were then placed in a
12.9 MHz nuclear magnetic resonance (NMR) rock core analyzer (Oxford Instruments) to
 SCA2016-016 7/12

enable self-diffusion measurements via the application of pulsed field gradient stimulated-
echo (PGSTE) techniques (Stejskal and Tanner, 1965). Measurement of the water’s self-
diffusion coefficient for a free liquid sample (D0) and for the water in the saturated rock
core (D), enables calculation of the system tortuosity (τ = D0/D) if a sufficiently long
observation time is allowed to ensure the diffusion within the core is completely restricted.
The same experiments were also run with methane at 3MPa providing a tortuosity
consistent with the value obtained from water. In future, tortuosity will be measured as a
function of the residual water content; D2O will be adopted to avoid NMR signal from this
wetting phase.

RESULTS AND DISCUSSION

‘Dry’ Sandstones and Carbonates
Table 2 summarizes all our experimental conditions (T, P and um), and a selection of
dispersion coefficients (Kcorr) for all the cores considered. The values of Kcorr have been
corrected for entry/exit and tubing effects as outlined above.

Core T (°C) P (MPa) u (mm s-1) D (10-8 m2 s-1) Kcorr (10-8 m2 s-1)
40 8 0.025 16.4 6.63
40 8 0.499 16.4 22.12
40 10 0.006 12.5 3.98
40 10 0.140 12.5 9.58
40 12 0.010 10.0 4.00
40 12 0.199 10.0 10.37
60 8 0.452 19.6 20.24
60 10 0.012 15.2 5.39
60 10 0.304 15.2 13.63
60 12 0.004 12.3 3.21
Berea
60 12 0.035 12.3 5.36
80 8 0.034 23.1 8.00
80 8 0.216 23.1 13.27
80 10 0.016 18.0 6.18
80 10 0.085 18.0 8.47
80 12 0.120 14.7 8.45
80 12 0.482 14.7 18.67
100 8 0.085 26.6 10.17
100 10 0.085 21.0 8.45
100 10 0.142 21.0 9.64
100 12 0.085 17.2 7.81
100 12 0.207 17.2 10.18
40 8 0.025 16.4 7.21
40 8 0.140 16.4 18.22
Donnybrook
40 12 0.050 10.0 6.69
40 12 0.141 10.0 18.66
 SCA2016-016 8/12

60 10 0.007 15.2 5.17

60 10 0.304 15.2 39.95
60 12 0.008 12.3 2.72
60 12 0.035 12.3 5.36
40 8 0.09 16.4 10.38
40 10 0.004 12.5 3.79
40 10 0.04 12.5 5.76
40 12 1.151 10 140.03
Estillades
60 10 0.23 15.2 21.59
60 10 1.149 15.2 133.65
60 12 0.016 12.3 4.39
60 12 1.646 12.3 214.59
40 8 0.053 16.4 17.08
40 10 0.011 12.5 8.49
40 10 0.141 12.5 27.06
40 10 1.124 12.5 212.49
Ketton
40 12 1.743 10 381.7
40 14 1.908 8.1 424.03
60 10 0.015 15.2 10.03
60 10 0.745 15.2 115.94
Table 2 Dispersion coefficients of equimolar mixtures of CO2–CH4 into pure CH4 or pure CO2 for Berea,
Donnybrook, Estillades and Ketton cores.

Application of Eq.(2) allowed the dispersivity (α) to be determined for the two sandstone
cores. For the Berea and Donnybrook rocks, α = 0.35 mm and 1.31 mm, respectively, with
a relative statistical uncertainty in these parameters from the regression of less than 5%.
These values are consistent with those reported for sandstones by Coats et al. (2009b),
although in the case of the Berea sandstone, the dispersivity measured here is moderately
lower than other values reported in the literature (1.2 mm (Gist et al. (1990)); 3.75 mm
(Brigham et al. (1961)); 2.2 mm (Legatski and Katz (1967)); 1-6 mm (Schulze-Makuch
(2005))). However, as demonstrated by the discussions above, the value of the apparent
dispersivity obtained can readily be affected by systematic measurement errors,
compounded by the effects on α due to variation in sample length employed and, of
course, variability in the actual samples used.

In Figure 3 the pulse breakthrough profiles of Estillades and Ketton carbonates show the
reproducibility of the data and the better fitting obtained for Ketton with the MIM model in
comparison to the AD equation.
 SCA2016-016 9/12

Figure 3 Left: repeatability of the pulse-breakthrough profiles for Estaillades carbonate core at 60◦C, 10 MPa
and 0.14 mm s−1. Right: Ketton carbonate core at 60◦C, 10 MPa and 0.28 mm s−1. The lower panel
demonstrates the breakthrough profiles and the fits achieved using Eq. (3) for Estaillades and both Eqs. (3)
and (4) for Ketton. The upper panel shows the resulting residuals of the fits.

Our previous results showed the 1D-AD equation was adequate for the description of
dispersion of CO2-CH4 system through sandstone cores. Larger dispersion coefficients at
similar conditions for Donnybrook sandstone were reported, indicating a higher level of
heterogeneity relative to the homogenous Berea sandstone. Because carbonate rocks are
known as heterogeneous porous media, long tailing and early breakthrough profiles of the
CO2 together with comparatively higher dispersion coefficients were expected. After
comparing several measured dispersion coefficients of the sandstones and carbonates at
similar conditions, the dispersion coefficients for both carbonate samples were found to be
larger than those for the homogeneous sandstone rock (Berea) whereas Estaillades showed
similar dispersive behaviour to the more heterogeneous (Donnybrook) sandstone. Ketton
exhibited the greatest dispersion among all the core samples used during this study. The
pulse breakthrough profiles we observed for Ketton carbonate showed a more persistent
long tail and an earlier breakthrough than for Estaillades carbonate.

The calculated dispersivities were then used to produce Figure 4 where the ratios of
dispersion and diffusion coefficients, Kcorr/D, were plotted against Pem. Also shown are
two curves utilizing Eq.(2) where n was 1 for Pem<1 and 1.2 and 1.4 (power-law scaling)
for Pem>1 (Bijeljic and Blunt, 2006; Bijeljic et al., 2011). Figure 4 explicitly shows that
the curves produced using Eq.(2) with n = 1.2 for sandstones and n = 1.4 for carbonates are
an excellent fit with the Donnybrook and Ketton data respectively for Pem>10. This
provides experimental validation for the power law exponents derived from transport
simulations (e.g. Bijeljic and Blunt (2006) and Bijeljic et al. (2011)). Finally, the Pem<100
considered in the study is lower compared to other standard curves reported in literature
 SCA2016-016 10/12

(e.g. Seymour and Callaghan (1997)) since this the IR spectrometer could not detect the
breakthrough pulse for higher velocities due to the limited time resolutions of the
instrument.

Figure 4 Measured Kcorr/D vs. Pem for Berea and Donnybrook sandstone and Ketton and Estaillades
carbonate cores. The dashed lines indicate the independent NMR measurements of sandstone rocks
tortuosity. The fitted curves are produced by Eq. (2) for n equal to 1.2 and 1.4.

Inclusion of Residual Water

Figure 5 shows the impact of connate or residual water on the breakthrough profiles for
Donnybrook. The presence of this second phase clearly increases the dispersion
significantly, the corresponding dispersion coefficients are 5.7×10-7 and 39.0×10-7 m2·s-1
respectively. Future work will quantitatively measure the dependence of medium
properties such as dispersivity on the residual water content.

Figure 5 Pulse breakthrough profiles obtained for Donnybrook dry and 58% saturated with water. The
measurements were made at 40 °C, 10 MPa and with an interstitial velocities 0.34 mm·s-1.
 SCA2016-016 11/12

Measurement of Tortuosity
The NMR measurements of tortuosity (τ), using water with diffusion times between 10ms
and 1s, were 4.3 and 4.8 for Berea and Donnybrook sandstone cores, respectively. In
literature, lower values of tortuosities have been reported from electrical measurements
(Zhan et al. (2011)). The tortuosity value for the Berea rock core was also calculated by
measuring the self-diffusion coefficient of methane at 3MPa; the tortuosity value in this
case was 4.5 which is broadly in agreement with the water measurement. In Figure 6 we
show the raw signal attenuation NMR data – the slope of these data provides the diffusion
coefficient according to the Stejskal-Tanner equation (Stejskal and Tanner, 1965) and the
ratio of the free to restricted diffusion coefficient provides the tortuosity, τ. The diffusion
time for these measurements was 100 ms, ensuring the diffusion to be fully restricted
((Hurlimann (1994). In future we will measure τ as a function of water content.

Figure 6 NMR diffusion measurements of free and restricted methane at 3MPa.

CONCLUSIONS
A pulsed measurement apparatus was established for the measurement of the dispersion
coefficient (K) for SCCO2 and CH4 and successfully applied to two sandstones and two
carbonates samples. The methodology was adapted to minimize the effects of density and
concentration contrasts; the reproducibility obtained were excellent. By using 1D- ADE
approach for sandstones and Estillades carbonate and a mobile-immobile model to analyse
the breakthrough profiles observed for the Ketton carbonate, asymptotic dispersion
coefficients were obtained and compared. All acquired data collapsed onto a common
curve when K/D was plotted as a function of Pem, capturing variation with temperature and
pressure. Preliminary results showed the impact of residual water on the dispersion
mechanism and independent measurements of tortuosity with water and methane also
produced consistent results.

ACKNOWLEDGEMENTS
This work was supported by the Western Australian Department of Environment and
Conservation through the Low Emissions Energy Development fund.
 SCA2016-016 12/12

REFERENCES
Arya, A., Hewett, T. A., Larson, R. G., Lake, L. W., 1988. Dispersion and reservoir heterogeneity SPE Reservoir
Engineering 3, 139-148.
Berkowitz, B., Cortis, A., Dentz, M., Scher, H., 2006. Modeling non-fickian transport in geological formations as a
continuous time random walk, Reviews of Geophysics 44, 1-49.
Bijeljic, B., Blunt, M. J., 2006. Pore-scale modeling and continuous time random walk analysis of dispersion in
porous media, Water Resources Research 42, W01202.
Bijeljic, B., Mostaghimi, P., Blunt, M. J., 2011. Signature of non-Fickian solute transport in complex
heterogeneous porous media, Physical Review Letters 107, 204502.
Brigham, W. E., Reed, P. W., Dew, J. N., 1961. Experiments on mixing during miscible displacement in porous
media, SPE Journal 1, 1-8.
Coats, K. H., Smith, B. D., 1964. Dead-end pore volume and dispersion in porous media, SPE Journal 4, 73-84.
Coats, K. H., Whitson, C. H., Thomas, L. K., 2009a. Modeling Conformance as Dispersion, SPE Reservoir
Evaluation & Engineering 12, 33-47.
Coats, K. H., Whitson, C. H., Thomas, L. K., 2009b. Modeling conformance as dispersion, SPE Reservoir
Evaluation & Engineering 12, 33-47.
Deans, H. A., 1963. A mathematical model for dispersion in the direction of flow in porous media Society of
Petroleum Engineers Journal 3, 49-52.
Gist, G. A., Thompson, A. H., Katz, A. J., Higgins, R. L., 1990. Hydrodynamic dispersion and pore geometry in
consolidated rock, Physics of Fluids A 2, 1533-1544.
Haggerty, R., Gorelick, S. M., 1995. Multiple-rate mass transfer for modeling diffusion and surface reactions in
media with pore-scale heterogeneity, Water Resources Research 31, 2383-2400.
Hughes, T. J., Honari, A., Graham, B. F., Chauhan, A. S., Johns, M. L., May, E. F., 2012. CO2 sequestration for
enhanced gas recovery: new measurements of supercritical CO2-CH4 dispersion in porous media and a
review of recent research, International Journal of Greenhouse Gas Control 9, 457-468.
Hurlimann, M. D., Helmer, K. G., Latour, L. L., Sotak, C. H., 1994. Restricted diffusion in sedimentary rocks.
determination of surface-area-to-volume ratio and surface relaxivity, Journal of Magnetic Resounance,
Series A 111, 169-178.
Lake, L. W., 1989. Enhanced oil recovery Prentice Hall 550, New Jersey, USA.
Legatski, M. W., Katz, D. L., 1967. Dispersion coefficients for gases flowing in consolidated porous media, SPE
Journal 7, 43-53.
Levenspiel, O., 1999. Chemical reaction engineering 3rd ed. John Wiley & Sons.
Perkins, T. K., Johnston, O. C., 1963. A review of diffusion and dispersion in porous media, SPE Journal 3, 70-84.
Pooladi-Darvish, M., Hong, H., Theys, S., Stocker, R., Bachu, S., Dashtgard, S., 2008. CO2 injection for enhanced
gas recovery and geological storage of CO2 in the long Coulee Glauconite F pool, Alberta. In SPE (Ed.),
SPE Annual Technical Conference and Exhibition, Denver, Colorado.
Schulze-Makuch, D., 2005. Longitudinal dispersivity data and implications for scaling behavior, Ground Water 43,
443-456.
Seymour, J. D., Callaghan, P. T., 1997. Generalized Approach to NMR Analysis of Flow and Dispersion in Porous
Media, AIChE Journal 43, 2096-2111.
Stejskal, E. O., Tanner, J. E., 1965. Spin diffusion measurements: spin echoes in the presence of a time dependent
field gradient The Journal of Chemical Physics 42, 288-292.
Takahashi, S., Iwasaki, H., 1970. The diffusion of gases at high pressures. III. The diffusion of 14CO2, in the 12CO2-
CH4 system, Bulletin of the Chemical Research Institute of Non-Aqueous Solutions, Tohoku University
20, 27-36.
Van Genuchten, M. T., Wierenga, P. J., 1976. Mass transfer studies in sorbing porous media I. analytical solutions,
Soil Science Society of America Journal 40, 473-480.
Vandeweijer, V., van der Meer, L. G. H., Hofstee, C., Mulders, F., Graven, H., D'Hoore, D., 2011. Monitoring CO2
Injection at K12-B http://www.co2geonet.com/UserFiles/file/Open%20Forum%202011/PDF-
presentations/2-10_Vanderweijer.pdf.
Zhan, X., Zhu, Z., ToksoZ, M. N., 2011. Quantitative DC and high frequency AC seismoelectric measurement on
Berea sandstone, SEG Technical Program Expanded Abstracts 30, 2246-2250.